Application Type NDA Submission Number 50-679 Reviewer Name

CLINICAL REVIEW

Application Type NDA

Submission Number 50-679

Reviewer Name Peter Kim, MD, MS

Review Completion Date 1/20/09, revised 4/6/09

Established Name Cefepime

Trade Name Maxipime

Therapeutic Class Cephalosporin (4th generation)

Applicant Bristol-Myers Squibb (BMS)

Formulation Parenteral

Dosing Regimen Variable (from 500 mg q12h to 2 gm q8h)

Indications Pneumonia, Empiric Therapy for Febrile Neutropenic Patients, Uncomplicated and Complicated Urinary Tract Infections (including pyelonephritis), Uncomplicated Skin and Skin Structure Infections, Complicated Intra-abdominal Infections

Intended Population Patients aged > 2 months

Table of Contents 1 EXECUTIVE SUMMARY ...... 3 1.1 RECOMMENDATION ON REGULATORY ACTION ...... 3 1.2 RECOMMENDATION ON POSTMARKETING ACTIONS ...... 3 1.2.1 Risk Management Activity...... 3 1.2.2 Post marketing requirements...... 4 1.2.3 Other Phase 4 Requests...... 4 1.3 SUMMARY OF CLINICAL FINDINGS...... 4 1.3.1 Brief Overview of Clinical Issue...... 4 1.3.2 Efficacy ...... 4 1.3.3 Safety ...... 5 2 INTRODUCTION AND BACKGROUND...... 6 2.1 REGULATORY ACTIVITY (ABRIDGED VERSION)...... 6 2.2 OTHER RELEVANT BACKGROUND INFORMATION ...... 14 3 DATA SOURCES, REVIEW STRATEGY, AND DATA INTEGRITY ...... 26 3.1 REVIEW STRATEGY ...... 26 3.2 THE YAHAV ET AL. PAPER...... 29 3.3 CHARACTERIZATION OF THE BMS DATA REPOSITORY FOR CEFEPIME CLINICAL TRIALS ...... 33 4 INTEGRATED REVIEW OF SAFETY...... 42 4.1 METHODS AND FINDINGS ...... 42 4.1.1 Deaths ...... 50 4.111 Collaborative Work with Ana Szarfman, M.D., Ph.D...... 140 4.112 Highlights from the analysis performed by Yu-te Wu, Ph.D. (Statistical reviewer, Quantitative Safety and Pharmacoepidemiology group) ...... 186 5 ADDITIONAL CLINICAL ISSUES...... 220 5.1 SANZ ET AL. (2002)...... 220 5.2 CURRENT CEFEPIME SUSCEPTIBILITY ...... 225 5.3 MEDICAL OFFICER DISCUSSION ...... 226 6 OVERALL ASSESSMENT ...... 230 6.1 CONCLUSIONS ...... 230 6.2 RECOMMENDATION ON REGULATORY ACTION ...... 230 6.3 RECOMMENDATION ON POSTMARKETING ACTIONS ...... 230 6.3.1 Risk Management Activity...... 230 6.3.2 Post marketing requirements...... 231 6.3.3 Other Phase 4 Requests...... 231 6.4 LABELING REVIEW...... 231 6.5 COMMENTS TO APPLICANT ...... 232 7 APPENDICES...... 233 7.1 REGULATORY ACTIVITY (UNABRIDGED VERSION WITH M.O. COMMENTS) ...... 233 7.2 HIGHLIGHTS FROM THE ORIGINAL MO REVIEW OF BMS’S FIRST CEFEPIME NDA (COMPLETED IN 1994) .....254 7.3 ADDITIONAL INFORMATION FROM THE 1997 JOINT CLINICAL AND STATISTICAL REVIEW OF CEFEPIME AS EMPIRIC MONOTHERAPY FOR FEBRILE NEUTROPENIA ...... 256 7.4 CEFEPIME CLINICAL TRIALS INCLUDED IN THE META-ANALYSIS BY YU-TE WU, PH.D...... 302 7.5 LINE-BY-LINE LABELING REVIEW ...... 305 7.6 CLINICAL REVIEW OF BMS’ SUBMISSION ENTITLED, “A META-ANALYSIS OF BMS RANDOMIZED CEFEPIME TRIALS”...... 306 8 REFERENCES ...... 328

1 EXECUTIVE SUMMARY

1.1 Recommendation on Regulatory Action

Due to its antibacterial activity against Gram-positive and Gram-negative pathogens, including Pseudomonas aeruginosa and Enterobacteriaceae; the lack of a clear, biologically plausible reason for the increased mortality observed in the meta-analysis by Yahav et al.1 [Lancet Infect Dis. 2007 May;7(5):338-48]; and the lack of a statistically significant mortality difference in the Agency’s meta-analysis (which included the studies in the Yahav meta-analysis plus additional studies not included in the Yahav analysis), the M.O. currently recommends that cefepime remain on the market and maintain all approved indications. The M.O. notes that only adequately powered and well-controlled prospective trials may definitively answer the question as to whether or not the use of cefepime is associated with increased mortality. However, given the practical limitations of executing such trials using a mortality endpoint, and in particular, additional trials to study cefepime as empiric monotherapy in febrile neutropenic patients, the following additional risk management activities will be recommended.

1.2 Recommendation on Postmarketing Actions

1.2.1 RISK MANAGEMENT ACTIVITY

The M.O. recommends that the following risk management activities move forward. 1. The Agency, through the Office of Surveillance and Epidemiology (OSE), in conjunction with the Premier Healthcare Informatics database, is poised to perform a postmarketing analysis of mortality associated with the administration of cefepime versus comparable agents. 2. BMS, in conjunction with the (b) (4) Drug Utilization Database, is also poised to perform their own postmarketing analysis of mortality associated with the administration of cefepime versus comparable agents. 3. The following issue is currently under review. The Division may consider changing the dosing recommendations and/or mean inhibitory concentration (MIC) susceptibility breakpoints for P. aeruginosa and the Enterobacteriaceae. This inclusion of additional language in the “Microbiology” section of the U.S. label would improve consistency between the U.S. and EMEA product labels due to a change in bacterial resistance to cefepime (as well as other antibacterial agents). 4. The Division may consider providing additional clarification in the INDICATIONS/USAGE section of the label for “Empiric Therapy for Febrile Neutropenic Patients”. The proposed language is discussed in Section 6.4 of this review.

In the event that BMS chooses to perform additional prospective studies using cefepime in the future, both BMS and Agency should continue to evaluate for mortality imbalances against cefepime.

1.2.2 POST MARKETING REQUIREMENTS

The M.O. recommends that BMS perform ongoing postmarketing analyses of mortality associated with cefepime use. This may include using large hospital databases to perform observational analyses of mortality for patients treated with cefepime versus comparable antibacterial agents.

1.2.3 OTHER PHASE 4 REQUESTS

The Agency asked BMS to perform their own meta-analysis to assess mortality associated with the use of cefepime versus comparable agents. (Results are found in Appendix 7.6 of this review.)

1.3 Summary of Clinical Findings

1.3.1 BRIEF OVERVIEW OF CLINICAL ISSUE

On May 16, 1997, the Agency approved cefepime for use as empiric therapy in febrile neutropenic patients. Cefepime remains the only antibacterial agent approved as monotherapy for this indication and is currently listed as one of the first line empiric therapies for febrile neutropenia by the Infectious Diseases Society of America (IDSA).2 In May 2007, Yahav et al.1 published a meta-analysis in The Lancet Infectious Diseases that noted an increased 30-day mortality associated with the use of cefepime versus other β-lactams for the following combined clinical conditions: neutropenic fever, pneumonia, urinary tract/gynecologic infections, and other/mixed infections. However, upon further review of the paper, the increased risk of death was only significant for the clinical condition of febrile neutropenia, and the authors were not able to provide a biologically plausible explanation for this apparent increased risk of death. This publication sparked further interest in a possible association between cefepime use and increased mortality. The Agency opened a dialogue with the holder of the reference listed drug, Bristol-Myers Squibb (BMS), in July 2007 to further investigate this issue.

The purpose of the current review was to investigate whether a true mortality imbalance exists against cefepime versus other β-lactam agents overall, and specifically, for the indication of febrile neutropenia. If such an imbalance was found, then the plan was to determine a biologically plausible explanation for this increased mortality.

1.3.2 EFFICACY

As the purpose of this review was to determine whether a true mortality imbalance exists against cefepime versus other β-lactam agents overall, and specifically, for the indication of febrile neutropenia, efficacy was only addressed from the perspective of safety, e.g., a lack of cefepime efficacy against specifically labeled pathogens would be considered a safety issue.

1.3.3 SAFETY

The purpose of this review was to determine whether use of cefepime is associated with an increased risk of mortality relative to comparable β-lactam antibacterial agents, with a focus on febrile neutropenia. The most straight-forward hypothesis for the mortality imbalance was that appropriate randomization was not obtained and more severely ill patients were enrolled in the cefepime arm of some studies, in particular, some that studied febrile neutropenia. However, additional hypotheses were evaluated, including: (1) a search for previously unidentified adverse event(s), such as, a neurological effect leading to death from undiagnosed seizures/status epilepticus, and (2) a lack of microbiological efficacy. In addition, meta-analytic methodology and WebSDM sector maps of adverse events were utilized to attempt to identify any other possible cefepime-related adverse event(s) that might be linked to excess mortality.

During the course of the review, the Agency has been evaluating more recent cefepime susceptibility data, particularly against Pseudomonas aeruginosa and Enterobacteriaceae. Though this was not found to be a key factor in the increased mortality associated with cefepime use in the nine febrile neutropenia studies previously submitted to the Agency for registration, this may require additional attention based on more current preliminary susceptibility data. Such measures may include a change in the dosing recommendations and/or lowering the susceptibility breakpoints for cefepime against Pseudomonas aeruginosa and Enterobacteriaceae. It should also be noted that (b) (5)

. This issue is currently under review by the Agency.

2 INTRODUCTION AND BACKGROUND

Currently, there are two generic forms of cefepime IV on the market in the U.S.: (1) ANDA 065 369 by Orgenus Pharma Inc. (U.S. Agent for Orchid Healthcare) approved 6/18/07, and (2) ANDA 065-441 by ACS DOBFAR approved 3/19/08. Additionally, Baxter Healthcare Corporation received approval for NDA 050-817 on 8/5/2008 for cefepime in the GALAXY plastic container.

2.1 Regulatory Activity (Abridged Version)

Please refer to Appendix 7.1 for an unabridged version of the Regulatory Activity with additional Medical Officer comments.

June 30, 1992 Original NDA 50,679 was submitted. It contained two studies (AI411 118 and -131) in support of efficacy for use as empiric therapy in febrile neutropenia; and one additional noncomparative study (AI411-143) in support of safety for the proposed indication.

January 4, 1994 Clinical review of the original NDA 50,679 by William Erhardt, M.D. found: AI411-118 - Equivalence not demonstrated between cefepime and control for empiric therapy of febrile neutropenia. AI411-131 - Equivalence demonstrated between cefepime and control for empiric therapy of febrile neutropenia but with only 22 patients in cefepime arm; majority of patients at one center. AI411-143 (open, noncomparative study) - Clinical response rate similar to comparative studies.

July 26, 1994 The FDA informed BMS that the indication of febrile neutropenia had been found to be non-approvable; proposed requirements for approval included either: One adequate, well-controlled study with evidence of effectiveness in at least three of the following: nosocomial pneumonia, complicated intra- abdominal infections, complicated skin and soft tissue infections, acute osteomyelitis, and acute bacterial arthritis or Appropriate results from two independent, adequate and well-controlled Febrile Neutropenia studies.

January 18, 1996 BMS received approval for the original cefepime NDA for Pneumonia (moderate to severe); Uncomplicated and Complicated Urinary Tract Infections (including pyelonephritis); Uncomplicated Skin and Skin Structure Infections; and Complicated Intra-abdominal Infections (used in combination with metronidazole).

May 17, 1996 Supplement SE1-002 to NDA 50,679 submitted; contained nine efficacy studies (5 controlled monotherapy, 2 controlled combination therapy, and 2 non-comparative studies) in support of an indication for use as empiric therapy in febrile neutropenic patients. Proposed labeling: (b) (4)

March 5, 1997 Presentation to the Anti-Infective Drug Products Advisory Committee (AIDPAC) of the preliminary FDA analysis of the five trials comparing cefepime monotherapy to either ceftazidime monotherapy or combination therapy, along with presentations by BMS of their analysis and by FDA consultants on scientific and clinical trial design issues relevant to the indication of febrile neutropenia. The AC discussed considerations related to the use of antibiotics for this indication and voted 7-0 that the data and analyses as presented supported the claim of safety and effectiveness of Maxipime® (cefepime) for empiric therapy of febrile neutropenic patients.

May 16, 1997 The Agency notified BMS that their supplemental application provided adequate evidence “for the empiric use of Maxipime® as monotherapy in febrile neutropenic patients”.

May 2007 Yahav et al. published a meta-analysis entitled, “Efficacy and safety of cefepime: a systematic review and meta-analysis”, in Lancet Infect Dis. 2007 May;7(5):338-48. In a meta-analysis of 57 trials, all-cause mortality was higher with cefepime as compared with other beta-lactams (risk ratio [RR] 1.26 [95% CI 1.08-1.49]).

M.O. comment: The Yahav et al. meta-analysis will be discussed in further detail later in this review.

July 25, 2007 The Division requested that BMS provide interpretation and comments regarding Yahav et al.1’s findings, as well as any necessary changes to the product labeling related to this issue.

August 15, 2007 The Division requested the following information from BMS: (1) mortality based on microbiologic data; (2) mortality by indication; (3) mortality stratified by monotherapy versus combination therapy; (4) mortality based on renal function; (5) mortality based on length of treatment; (6) timing and cause of death; (7) summary of mortality data from unpublished studies; (8) summary of data available on neurological adverse reactions associated with cefepime; (9) copies of all safety summaries previously submitted to NDA 50-679.

M.O. comment: The “Unpublished studies” included those studies that were not originally submitted to the U.S. FDA for registrational purposes. The actual number of “Unpublished studies” changed several times over the course of a year as BMS continued to review their trial databases. As of 9/8/08, BMS noted 51 “Unpublished studies”.

October 5, 2007 In addition to revisions to 8/30/07 responses, BMS provided summary data on neurological adverse reactions associated with cefepime.

October 26, 2007 BMS provided additional analyses as requested by the Division on 8/15/07: (1) mortality presented by microbiologic data; (2) mortality based on renal function; (3) mortality based on length of treatment; (4) timing and cause of death; (5) summary of mortality data from unpublished studies; and death narratives for all cefepime patients who were enrolled in BMS-sponsored studies. BMS noted that none of the submitted analyses pointed to a cefepime-related safety signal.

November 7, 2007 The Division made the following information request. “In the meta-analysis published in Lancet Infectious Diseases, May 2007, mortality data were available from 41/57 studies (63% of the population). Does the Sponsor have mortality data on any of the 16 studies for which data were not available in the published literature?”

On 12/3/07, BMS responded and noted that 9/16 “missing studies” were sponsored by BMS and that information from these studies was submitted in the 10/26/07 submission. On 12/14/07, the Division requested the following additional information. “Regarding your 12/3/07 submission related to the mortality data on 9 out of the 16 missing studies, we were unable to find the mortality data for 3 of the 9 studies reported to have such mortality data. Please provide the mortality data (total number of patients in each treatment arm and the number of patients who died per treatment arm) for the following studies. (The Division notes that BMS references their response submitted to the FDA on 10/26/07, however we have been unable to find the data of interest.) CPM 23-93.002, Jehn et al. 1998 (febrile neutropenia) AI411-174, Saito et al. 1992 (pneumonia study) AI411-168, Saito et al. 1992 (chronic respiratory tract infections)”

BMS responded on 12/16/07 and provided explicit mortality data for the three studies in question.

November 14, 2007 The Agency released the following “Early Communication About an Ongoing Safety Review Cefepime (marketed as Maxipime)”.19

December 10, 2007 The Division requested the following. “We acknowledge your response from 10/26/07, however, in light of the recent findings by Bhat et. al. (1), please analyze all mortality data (since the initial approval of cefepime) stratified by: (A) entry minimum inhibitory concentrations (MICs) for causative pathogens [that is, MIC < 1mg/L, 2 mg/L, 4

mg/L, 8 mg/L, and > 16 mg/L] and (B) treatment indication. Provide day of death relative to start of cefepime therapy and baseline APACHE II score, if available. Reference Cited: (1) Bhat SV, Peleg AY, Lodise TP Jr, Shutt KA, Capitano B, Potoski BA, Paterson DL. Failure of current cefepime breakpoints to predict clinical outcomes of bacteremia caused by gram-negative organisms. Antimicrob Agents Chemother. 2007 Dec;51(12):4390-5. Epub 2007 Oct 15.”3

January 7, 2008 The Division requested the following. “BMS has submitted narratives for all cefepime patients who died while participating in clinical studies. Please submit narratives for all comparator patients who died while participating in the same clinical studies. For the purposes of prioritization, send the death narratives for comparator patients in the febrile neutropenia studies first, and within the next week, if possible.”

Between 1/14 and 1/15/08, BMS provided the death narratives for comparator patients from the following clinical study programs: (1) febrile neutropenia, (2) complicated intra-abdominal infections, (3) the original NDA, (4) the updated integrated summary of safety from 1995, and (5) the pediatric studies.

January 8, 2008 BMS in conjunction with the Swiss drug regulatory agency, Swissmedic, published an “Important Information for Health Care Professionals” letter concerning, “Results from a meta-analysis publication reporting increased mortality associated with the use of cefepime.”

January 23, 2008 The Division requested the following. “In a communication dated 12/3/07 from BMS to the Division, BMS noted that mortality data were not available for the following study of febrile neutropenic patients by Sanz et al (2002) in the Journal of Antimicrobial Chemotherapy. However, the Division notes on page 87 of the attached paper, under the "Acknowledgements" section, that: "This work was supported in part by a grant from Bristol-Myers Squibb, Princeton, NJ, USA…"

“Therefore, the Division requests that BMS obtain and provide the necessary mortality data (total number of patients in each treatment arm and the number of patients who died per treatment arm).”

BMS responded on 2/1/08 with the following. “As noted in our previous response of 3 December 2007, this study was an Investigator Sponsored Trial (IST) and was not sponsored by BMS. As was common practice at the time this study was conducted, BMS routinely gave grants to investigators that applied for them in order to conduct clinical research. BMS was not involved in the conduct of these trials, accepted no responsibility for initiation, enrollment, safety reporting, data collection and analysis, or any other aspect of these trials and, as a result, the data from these trials belong

solely to the investigators. We have no access to the data beyond that which was made publicly available in the publication referenced above.”

The Division was able to make initial contact with Dr. Sanz via e-mail; however he subsequently did not respond to multiple requests for primary data on his study patients, e.g., causes of death for patients stratified by treatment group. Then, on 6/13/08, a colleague of Dr. Sanz, Dr. Isidro Jarque, responded with data on the causes of death for the clinically evaluable patients in the published febrile neutropenia study. Briefly, there were a total of 46 deaths out of 867 clinically evaluable patients (19/432 cefepime patients and 27/435 piperacillin/tazobactam patients).

M.O. comment: More detailed results from the Sanz et al. study will be provided in Section 5.1 of this review.

January 29, 2008 The Division requested the following. “Specify which studies in the published meta-analysis (1) are the studies for which BMS has mortality data. For example: The study by Kebudi, 2001 [ref 45 in the Yahav et al. (2007) paper] is derived from data from BMS study “AI411-…”. (1) Yahav et al. Efficacy and safety of cefepime: a systematic review and meta analysis. Lancet Infect Dis. 2007 May;7(5):338-48.”

BMS responded on 3/7/08 and noted that they sponsored 23 clinical studies that could account for 25 out of the 57 publications assessed by Yahav et al. (Out of these 23 clinical studies, 8 were not submitted to the U.S. FDA for NDA registration and were not included in BMS’s overall mortality calculation submitted to the FDA on 10/5/07.)

February 7, 2008 The Division requested the following. “Based on the revised "Table 1: Comparative Cefepime Trials by Indication" that was submitted by BMS on 10/5/07, there were 13 comparator patient deaths in Study AI411-204. However, the 1/14/08 submission of febrile neutropenia comparator patient death narratives only contains 10 death narratives for Study AI411-204. Please provide the 3 missing comparator patient death narratives from Study AI411-204.”

On the same day, the Applicant responded via e-mail that the revised table entitled, "Table 1: Comparative Cefepime Trials by Indication", submitted on 10/5/07, was incorrect and that there were actually only 10 comparator deaths (not 13) in Study AI411-204.

Also on 2/7/08, the Division requested the following. “Given that BMS reported conducting 45 comparative and non-comparative clinical trials of cefepime that included 5949 cefepime patients and 317 cefepime deaths (based on BMS submission from 10/5/07), explain the rationale behind what studies were provided to Dr. (b) (4) for his analysis (submitted to the Agency on 1/31/08).”

BMS provided the following response on 2/14/08:

“The request received from the Agency specified studies that were conducted "since the initial approval of cefepime", and thus all studies that were conducted as a part of the initial filing were excluded. In addition, many of the post marketing trials for our antibiotic products that were not planned to support label change filings did not capture MIC or other susceptibility data. Thus, only those trials that were used to support label changes post initial approval were included in the data sent to Dr. (b) (4) . Using these selection criteria, the package sent to Dr(b) (4) for his analysis was complete.”

Also on 2/7/08, the Division requested the following. “Provide the case report forms for all patients who died in both cefepime and comparator arms of cefepime studies (published and unpublished).”

BMS provided available CRFs in installments from 3/31/08 to 12/19/08.

February 11, 2008 The Division made the following request. “On January 6, 2008 the Agency requested that you review the interpretive criteria in the label for cefepime and provide supporting documentation to the Agency for review. In conjunction with the Agency’s request, the Division requests that you 1) provide recent MIC data for the organisms listed below, 2) provide pharmacokinetic parameters (e.g., plasma clearance, volume of distribution, elimination rate constant, and inter-compartmental constants) from healthy subjects from Phase 1 clinical trials using clinically-relevant doses, and 3) perform Monte Carlo simulations using the requested MIC and pharmacokinetic data to determine the probability of target attainment with a PK/PD target of 60-70% T>MIC for each organism listed under MIC Data…”

BMS provided their response on 9/29/08. In the response, BMS noted that they did not feel that a change in breakpoints was justified.

March 17, 2008 The Division requested a “Master Table” of all cefepime clinical trials conducted by BMS and non-BMS sponsors, design characteristics of the trials, and the mortality rates for cefepime and comparator. BMS provided an initial response to most of the questions on 7/31/08. However, the Agency noted inconsistencies which led to a teleconference with BMS on 8/15/08. This led to a revision of the BMS response on 9/8/08 to incorporate corrections to the data.

Also on 3/17/08, the Division requested patient level data from the febrile neutropenia datasets to further evaluate these patients for potential reasons for the mortality imbalance. BMS provided the requested datasets on 5/30/08; however, problems were noted which required resubmission of the datasets on 8/29/08.

March 25, 2008 The Division made the following request. “Based on BMS surveillance data since the original approval of cefepime on 1/18/1996, provide over progressive 2-3 year intervals, MIC ranges and MIC90s for: methicillin-susceptible Staphylococcus aureus, methicillin-resistant Staphylococcus aureus, Enterobacteriaceae (non-ESBL, and ESBL producers) (identify specific species), Pseudomonas aeruginosa,

Pseudomonas spp., (identify specific species) Acinetobacter spp., (identify specific species)”

On 11/14/08, BMS provided their response to this request.

April 9, 2008 The Division and Sponsor held a teleconference. BMS agreed to: (1) attempt to contact the authors of the publications included in the meta analysis that had missing mortality data; (2) provide further details about the 68 studies not submitted to the FDA (e.g., indications studied, mortality by treatment arm, and reasons for non-availability of data); (3) provide datasets for further analysis; and (4) conduct a meta-analysis that includes all publications that were included in the meta-analysis plus additional BMS sponsored studies that were not included in the meta analysis.

On 5/13/08, BMS attempted to provide additional clarification on the 57 references in the meta-analysis publication and the 68 non-registrational trials, including: (1) numbers enrolled per arm, (2) number of deaths per arm, (3) extent of available individual patient data, (4) explanation for why data were unavailable, e.g., author not responsive or BMS data not retained.

Based on further Agency inquiries, BMS provided additional clarifications on 7/31/08 and 9/8/08.

May 14, 2008 The Agency provided a “Follow-up to the November 14, 2007, Communication about the Ongoing Safety Review of Cefepime (marketed as Maxipime)”.20 The communication noted the following.

“…FDA is continuing to review safety data for the drug cefepime. There are a large number of studies to be reviewed and some, but not all, of the requested study data have been received by the FDA. As a result, FDA has not reached a definitive conclusion as to whether or not the increased mortality seen with cefepime compared to other β-lactam antibiotics observed in the meta-analysis is due to cefepime.”

June 7, 2008 The Division made the following request. “1. Clarify whether the following three patients enrolled in Febrile Neutropenia studies died within 30 days of study drug completion. If these patients did die within 30 days of study drug completion, please provide their death narratives and case report forms. 411204008352 411186008009 411186012003

2. Please explain why the pediatric patients enrolled in Study AI411-131 were not represented in the Febrile Neutropenia datasets submitted to the Agency on 5-30-08.”

BMS provided the requisite information on the 3 patients on 8/15/08 and 8/29/08. Per BMS, all three patients died > 30 days post last dose of study therapy.

June 9, 2008 The Division made the following request. 1. In addition to the 7/7/08 information request sent to BMS requesting clarification on whether three patients enrolled in Febrile Neutropenia studies (411204008352, 411186008009, and 411186012003) died within 30 days of study drug completion, please clarify whether the following additional three patients died within 30 days of study drug administration.

If these patients did die within 30 days of study drug completion, please provide relevant death narratives and case report forms…”

2. Regarding the 5/30/08 submission of datasets for the 7 comparative Febrile Neutropenia studies, and specifically the “Microbiology Susceptibility Test” Domain Dataset for each respective study, we note that the MSRESCAT (Result Category) variable only contains interpretive susceptibility information (‘Susceptible’, ‘Intermediate’, ‘Resistant’, ‘Null’) against cefepime and no interpretive susceptibility information for the comparator study drugs used in the 7 respective studies.

Provide a revised “Microbiology Susceptibility Test” Domain Dataset for each of the 7 comparative Febrile Neutropenia studies that contains interpretive susceptibility information (‘Susceptible’, ‘Intermediate’, ‘Resistant’, ‘Null’) against the respective comparator agents.

The revised Microbiologic Susceptibility data was provided on 8/29/08; and the requisite information on the three patients was provided in a submission on 9/2/08. Patient 411204005393 was later re-randomized to the ceftazidime group as 411204005471 and died <30 days post ceftazidime. Patient 411204016401 was later re-randomized to the ceftazidime group as 411204016484 and died <30 days post ceftazidime. Patient 411186036005 died > 30 days post therapy.

July 15, 2008 The Agency and Sponsor held a teleconference to discuss data elements in the Febrile Neutropenia dataset submitted 5-30-08.

July 31, 2008 BMS submitted: (1) the “Master Table” of all published and unpublished cefepime studies and (2) responses to several additional requests posed by the Division over the prior several months.

The “Master Table” of all the cefepime clinical studies was later revised and resubmitted by BMS on 9/8/08.

August 15, 2008 The Agency and BMS held a teleconference. The Agency requested that BMS: (1) revise the “Master Table” submitted on 7/31/08 to only include mortality data up to 30 days post therapy and if it was not possible to discern < 30 day from > 30 day mortality for some studies then to indicate those studies in the “Master Table” to the Agency; (2) clarify when mortality numbers provided by BMS in the “Master Table” differed from those provided in the Yahav et al. meta-analysis; and (3) clarify for febrile neutropenia studies whether data were based on numbers of patients or numbers of febrile neutropenia episodes and when the number of patients was not available to indicate those studies in the “Master Table”.

August 29, 2008 The Agency requested additional clarification on the meta-analysis datasets related to: (1) treatment assignment among comparator patients in Study 411-132, (2) the algorithm for how death status was established in the primary analysis dataset, and (3) how BMS coded patients without a final follow-up date or unknown date of death in the datasets.

BMS provided their responses on 9/10/08.

September 8, 2008 BMS submitted the revised “Master Table” of all clinical studies that contained a cefepime treatment arm.

September 25, 2008 BMS resubmitted the revised ‘define.xml’ files for the febrile neutropenia datasets.

September 29, 2008 BMS submitted their responses to Agency questions/comments related to cefepime MIC breakpoints and Monte Carlo simulations.

November 21, 2008 The Agency held a teleconference with BMS to further understand why more cefepime patients with resistant pathogens at baseline died as compared with comparator patients. BMS noted this may have been an artifact based on the fact that more comparator patients had missing information on baseline susceptibility to study therapy. In addition, BMS agreed to re-evaluate the integrity of their microbiology susceptibility datasets based on discrepancies noted by the Agency during their review of the originally submitted microbiology susceptibility data.

On 12/12/08, BMS submitted a revised microbiology susceptibility dataset and noted that their primary analysis dataset was “…updated to accommodate dirty/missing data as revealed by investigation of differences between PRIMANAL [primary analysis] and MS/MB [microbiology] datasets…”

December 30, 2008 BMS submitted their responses to Agency questions/comments related to BMS’s proposed postmarketing analysis of mortality associated with the administration of cefepime versus comparable agents using the (b) (4) Drug Utilization Database. BMS had originally submitted the proposal for a postmarketing analysis on 8/8/08. On 10/31/08, the Office of Surveillance and Epidemiology (OSE) provided comments/questions related to the proposed observational study.

January 22, 2009 The Agency and BMS held a teleconference to further clarify the protocol and statistical analysis plan. BMS stated that they would resubmit a revised protocol and statistical analysis plan.

2.2 Other Relevant Background Information

Cefepime was originally approved 1/18/1996 for the indications of Pneumonia (moderate to severe) caused by Streptococcus pneumoniae, including cases associated with concurrent

14 bacteremia, Pseudomonas aeruginosa, Klebsiella pneumoniae, or Enterobacter species; Uncomplicated and Complicated Urinary Tract Infections (including pyelonephritis) caused by Escherichia coli or Klebsiella pneumoniae, when the infection is severe, or caused by Escherichia coli, Klebsiella pneumoniae, or Proteus mirabilis, when the infection is mild to moderate, including cases associated with concurrent bacteremia with these microorganisms; Uncomplicated Skin and Skin Structure Infections caused by Staphylococcus aureus (methicillin-susceptible strains only) or Streptococcus pyogenes; and Complicated Intra- abdominal Infections (used in combination with metronidazole) caused by Escherichia coli, viridans group streptococci, Pseudomonas aeruginosa, Klebsiella pneumoniae, Enterobacter species, or Bacteroides fragilis.

Original NDA for Cefepime contained Febrile Neutropenia Studies The original NDA submission contained clinical trial data on the proposed indication of febrile neutropenia. However, the clinical reviewer of the original submission, Dr. William Erhardt, concluded that the data were inadequate to demonstrate that cefepime was equivalent to control regimens in empiric therapy for febrile neutropenic patients. The clinical reviewer drew the following conclusions on the three submitted febrile neutropenia studies. (1) In Study AI411 118, non-inferiority was not demonstrated between cefepime and control (piperacillin + gentamicin) for empiric therapy of febrile neutropenia. (2) In Study AI411-131, non-inferiority was demonstrated between cefepime and control (ceftazidime) for empiric therapy of febrile neutropenia but with only 22 patients in cefepime arm, and with the majority of patients at one center. (3) In Study AI411-143 (open, noncomparative study), the clinical response rate was similar to comparative studies. (4) Subgroup analysis suggested that cefepime monotherapy was inferior to control regimens in empiric therapy of febrile neutropenic patients. (5) The data suggested that cefepime may have had a safety advantage over aminoglycosides. BMS was informed of the conclusions in a teleconference and responded that they considered these data to be an “interim report” and that they planned to submit a supplemental (s)NDA for this indication by the end of 1994.

Please refer to Appendix 7.2 for additional highlights from the original M.O. review of BMS’s First Cefepime NDA (completed in 1994)

Supplemental NDA for Cefepime as Empiric Therapy for Febrile Neutropenic Patients The sNDA for cefepime as “empiric therapy for febrile neutropenic patients” was submitted to the Agency on 5/17/96. The “Joint Clinical/Statistical Review [JCSR] of NDA 50,679/SE1-002” was completed by David Ross, M.D., Ph.D. (clinical reviewer) and Aloka Chakravarty, Ph.D. (statistical reviewer) on 6/12/97. The JCSR noted that 9 studies were submitted: 3 studies compared cefepime and ceftazidime monotherapy (647 patients, comprising 733 episodes), 2 studies compared cefepime to a combination regimen of a β-lactam/aminoglycoside combination (187 patients comprising 187 episodes), 1 study compared cefepime in combination with vancomycin to the corresponding ceftazidime combination (111 patients comprising 128 episodes), and 1 study compared the combination of cefepime with amikacin to the corresponding ceftazidime combination (353 patients comprising 353 episodes). At the time the studies were conducted, there were no approved comparator regimens; however, all of the comparator regimens utilized in the cefepime development program were considered

15 contemporaneously clinically appropriate empiric therapy. Additionally, two non-comparative studies were also submitted.

M.O. comment: Given that BMS was unable to provide complete datasets for any other febrile neutropenia studies than those originally submitted to the sNDA, the M.O. reviewed the original findings of the JCSR and provided additional comment.

The JCSR contained the following introduction. The review of New Drug Applications (NDAs) for empiric therapy of febrile neutropenia is complicated by two factors. First, this indication has not previously been granted by the Division of Anti-Infective Drug Products (DAIDP); thus, there is currently no anti-infective agent approved for the indication of febrile neutropenia. Although the labeling of some antibiotics carries usage statements which imply that these drugs may be suitable for this indication, the language is non-specific and does not provide an adequate regulatory basis for identifying these drugs as suitable comparators for a controlled trial of empiric therapy. In general, the policy of the Center for Drug Evaluation and Research (CDER) is that when there is no FDA-approved agent for an indication, and an active agent is used as the comparator, the test agent must either 1) demonstrate statistical superiority to the active comparator for its efficacy to be established (i.e., the active comparator is treated as a placebo) or 2) demonstrate equivalence to an active comparator previously shown to be superior to placebo in the scientific literature. It is clear from the literature that empiric therapy of febrile neutropenia is superior to no therapy (Hathorn and Lyke 1997), and that the community standard of care is to employ empiric therapy in this setting. The sponsor has provided references in their submission to such literature. Thus, comparator regimens which are medically accepted may be considered valid as active controls for the purposes of reviewing the clinical trials contained in this submission. Following this logic, the test drug must show therapeutic equivalence to the comparator regimen to demonstrate efficacy, but statistical superiority is not required. The combination regimens employed in these trials include monotherapy with ceftazidime, combinations of an ureidopenicillin and an aminoglycoside, and combinations of ceftazidime with vancomycin or amikacin. All of these represent clinically accepted regimens for empiric therapy of febrile neutropenic patients (Pizzo et al. 1986; Hathorn and Lyke 1997), and therefore are acceptable from a regulatory standpoint as active comparators. However, it is important to note that there are subpopulations of febrile neutropenic patients who are at particularly high risk for severe infection, for whom some regimens, such as ceftazidime monotherapy, may not be appropriate. A second, and more fundamental problem, is the lack of consensus with respect to deciding which patients are evaluable for efficacy, and which endpoints are appropriate measures of outcome. These problems have complicated both clinical trial design and reporting of results (Pater and Weir 1986; Young 1988). M.O. comment: The problems outlined in the “Introduction” to the JCSR persist today: lack of consensus on an evaluable population and appropriate measures of outcome. In addition, due to the fact that neutropenic fever is more of a syndrome than an infectious disease, the patient population tends to be rather heterogeneous with regard to baseline co-morbid conditions and severity of illness. The major reason for this lack of consensus is that fever is not a specific marker of infection, yet is frequently the only marker available in this patient population. Patients who have a positive culture result or a clinically documented infection should generally be evaluable for efficacy; however, such patients may be in the minority in a clinical study of empiric therapy of febrile neutropenia. Many patients with neutropenia and fever will not have a clinically or microbiologically identifiable source of infection at presentation or at any point during their clinical course, and may in fact not have an infectious source of fever. A substantial number of these patients will defervesce only after resolution of agranulocytosis, making it difficult to determine whether antibiotics or bone marrow recovery were responsible for improvement. However, such patients cannot be identified a priori, and institution of antibiotic therapy only for microbiologically or clinically documented infections carries an unacceptably high risk of mortality from infection (Hathorn and Lyke 1997). M.O. comment: The M.O. notes that patients with a microbiologically or clinically documented infection are often in the minority in a clinical study of empiric antibacterial therapy for febrile neutropenia.

Thus, the clinical imperative that fever be treated promptly may make it difficult to conclude that an antibiotic regimen is ‘effective’ in treating infection, because it is not always clear that infection was present at the outset. However, fever cannot be disregarded as a clinical parameter, since neutropenic patients with fever must be presumed to be infected until proven otherwise. This paradox has been addressed in a variety of ways in the literature. Some authors (Pizzo et al. 1986) have advocated survival from infection as the primary outcome, since this is an incontrovertible measure of effectiveness. This approach does not consider modification of the original regimen to constitute failure, since prevention of early mortality is the major goal from this perspective. However, this approach does not take into account morbidity from persistent fever, including complications from repeated modifications of antibiotic regimens, and may obscure differences between antibiotic regimens (Elliott and Pater 1988). In addition, from a regulatory standpoint, this ap proach could potentially lead to regulatory decisions on empiric therapy for febrile neutropenia that are not based on clinical reality. For example, a febrile neutropenic patient who was treated empirically with penicillin and survived after addition of other antibiotics would be scored as a success under this approach. Thus, a trial of penicillin as empiric therapy that employed such a definition of success might show therapeutic equivalence between penicillin and a standard comparator regimen, supporting regulatory approval of penicillin for this indication; however, few clinicians, if any, would use penicillin for empiric therapy in this setting. The alternative approach that of considering defervescence as a primary outcome, is also problematic, for the reasons outlined above. In addition, this approach may lead to outcome decisions that are as paradoxical as the survival approach. For example, a patient who has fever alone, defervesces on empiric therapy, and subsequently develops an infection with an isolate with induced resistance to the original regimen would be considered a success, although from a clinical standpoint the patient may be worse off than if he had never received empiric therapy. The definition of primary outcome has a direct effect on which patients will be considered evaluable for efficacy. For example, if one considers defervescence as the main endpoint, with modification of the empiric antibiotic regimen considered a treatment failure, then patients should receive the initial regimen for a minimum duration of time in order to be evaluable for efficacy. The minimum length of time that would be required for evaluability may vary between different studies. If one is simply examining survival, however, then a minimum duration of therapy may not be needed. Similarly, protocol definitions of infections or allowable modifications of the anti-microbial regimen that may occur during empiric therapy affect patient evaluability. For example, it is not always clear whether neutropenic patients who have fever alone and receive both anti-bacterial agents and anti-fungal or anti-viral agents are evaluable, since such patients may have a non-bacterial infection. The Infectious Diseases Society of America has published guidelines for the evaluation of anti-infective drugs for the empiric treatment of febrile episodes in neutropenic patients that address these issues of clinical trial design and the principles for evaluation of response to therapy (Hughes et al. 1992). They provide important guidance on aspects of optimal study design, including study population selection, inclusion and exclusion criteria, comparator selection, allowable modifications, endpoints, and methods of data analysis. Although the IDSA guidelines are helpful in providing a framework for designing, performing, and analyzing studies in this area, many key issues remain unresolved from the regulatory standpoint. The questions of which primary and secondary endpoints should be emphasized in making regulatory decisions and which patients should be considered as evaluable in the review of NDAs for this indication remain incompletely answered. The Points to Consider (PTC) document issued by the Division of Anti-Infective Drug Products (DAIDP) does not explicitly address clinical trial designs and evaluability criteria, but it does suggest that for this indication, an application present data from at least one statistically adequate and well-controlled multicenter trial, along with previously established effectiveness for at least three specific deep infections (e.g., pneumonia, complicated urinary tract infections, complicated skin and skin structure infections). In addressing the problems described above, three goals were paramount. First, consistent evaluability and efficacy criteria were needed that would allow review of studies with similar trial designs, both in this application as well as in future applications which might be submitted to the FDA. Second, the data had to be analyzed to determine the efficacy and safety of Maxipime® for empiric therapy for febrile neutropenic patients, relative to a clinically acceptable comparator. Finally, if consistent with approval, the data and analysis had to form the basis for construction of a clinically useful and scientifically sound label. The strategy used to reach these goals centers around the observation by Hughes et al. (1992) that “… it is optimal to use multiple parameters for the assessment of [febrile neutropenic] patients, including clinical response to therapy, evidence of microbiologic efficacy, and survival.” In addition, as pointed out by Sackett and Gent (1979), the nature of an evaluable population and the endpoints used to analyze that population depend on the question being asked of the data and the necessity of “avoiding specific bias”. Thus, if one is asking whether an empiric regimen “works”, it is important to define what one is asking of the drug: Should it treat the initial infection, prevent subsequent infections (i.e., provide prophylaxis), or simply protect against early mortality from infection? Furthermore, should one examine efficacy in a well-defined population, or in a study population that mirrors that found in the real world?

The JCSR delineated the studies reviewed in the sNDA in the following manner. An overview of studies reviewed in this supplement is shown in Table 9.4. The studies fell into four groups: 1. Cefepime monotherapy compared to ceftazidime monotherapy (studies AI411-131, 189, and 204). 2. Cefepime monotherapy compared to a ß-lactam/aminoglycoside combination (studies AI411-118 and 137). 3. Cefepime in combination with amikacin (study AI411-186) or in combination with vancomycin (study AI411-198) compared with the corresponding ceftazidime combination. 4. Open-label, non-comparative studies (studies AI411-143 and 158). Studies were analyzed individually. In addition, studies AI411-204 and -189, the largest trials in Study Group 1 (studies comparing cefepime monotherapy and ceftazidime monotherapy) were also pooled for analysis. Trials in Study Groups 2, 3, and 4, comprising the studies comparing cefepime monotherapy to combination therapy (AI411 118, -137), those comparing cefepime combination therapy to ceftazidime combination therapy (AI411-186, -198), and the open-label studies (AI411-143, -158) were not pooled for efficacy analysis because of significant differences in trial design and heterogeneity of treatment effect (see below).

Table 9.4. Study Designs - NDA 50,679/SE1-0021

411-143 411-158 411-131 411-189 411-204 411-118 411-137 411-186 411-198

Type Phase II Phase II Phase III Phase III Phase III Phase III Phase III Phase III Phase III Design Open Open Open Open Double-blind Open Open Open Open Randomized Randomized Randomized Randomized Randomized Randomized Randomized Country/Continent Europe Netherlands U.S. Europe U.S. U.S. U.S. France Belgium Cefepime 2g q8h 2g q8h 2g q8h 2g q8h 2g q8h 2g q8h 2g q8h 2g q12h 2g q8h + +Amikacin Vancomycin Comparator N/A N/A Ceftazidime2g Ceftazidime2g Ceftazidime 2g Piperacillin/Gen Mezlocillin/Gen Ceftazidime2g Ceftazidime2g q8h q8h q8h tamicin tamicin q8h +Amikacin q8h + Vancomycin Start Date 1-90 2-91 8-89 2-93 1-93 6-89 7-90 10-92 2-93 End Date 6-91 11-91 11-91 6-95 11-94 12-91 8-92 11-93 2-94 No. of Sites 2 1 2 15 11 2 1 31 4 Accrual 84 30 90 281 276 116 71 353 111 Cefepime 84 30 45 139 143 59 35 242 53 Control ------45 142 133 57 36 111 58 Total Episodes 108 30 104 324 315 116 71 353 111 Cefepime 108 30 52 166 163 59 35 242 53 Control ------52 158 152 57 36 111 58

*Shaded columns represent studies submitted with original NDA. Study AI411-137 was submitted with the original NDA only as part of the safety database.

M.O. comment: The data submitted for Study 131 apparently only included those from the adult patients N=45 for cefepime and N=45 for ceftazidime. The pediatric data for this study was not apparently submitted for consideration in the sNDA for febrile neutropenia. It is also notable that among the Phase 3 studies, Studies 131, 189, and 204 had more episodes of febrile neutropenia than individual patients enrolled. This meant that patients with more than one episode of febrile neutropenia could be re-randomized to another course of empiric study therapy.

The JCSR Medical Officer noted the following on page 27 of the review. Although there is no FDA-approved comparator for this indication, ceftazidime has been shown to be effective as monotherapy for febrile neutropenic patients (Pizzo et al. 1986)4, and is accepted in the medical community for this purpose. However, the evidence that it is effective as monotherapy in patients at high risk for infection (e.g., patients with severe neutropenia) is weak (Sanders et al. 1991)5; results from clinical trials comparing different monotherapy regimens suggest that imipenem may be a more appropriate monotherapy comparator for high-risk patients (Liang et al. 1990)6.

M.O. comment: The M.O. notes that ceftazidime may also not be adequately effective as febrile neutropenia monotherapy for patients at high risk for infection.

The M.O. noted that in the paper by Pizzo et al. (1986), efficacy was assessed as response at 72 hours post initiation of therapy and at resolution of the neutropenia episode.

The two largest monotherapy trials, which were multi-center and randomized, were deemed by the JCSR reviewers to be poolable on the basis of similar study designs and homogeneity of treatment effect. For these pooled results, cefepime was considered: “…therapeutically equivalent to ceftazidime for all definitions of clinical response. However, the response rate for cefepime was lower than that for ceftazidime under all analyses, with the point estimate of the difference in the first episode response rates ranging from -4.3% (for the primary endpoint) to -3.4% for survival regardless of modification.”

M.O. comment: The M.O. notes that the original JCSR reviewers observed a survival difference of 3.4% against cefepime as compared with ceftazidime in the pooled studies (Studies 204 and 189).

Additionally, the JCSR noted the following. A subgroup analysis of the pooled monotherapy studies showed that response rates for cefepime were lower than those for ceftazidime for patients with the highest risk of having a severe bacterial infection (such as patients with an underlying hematologic malignancy, those with severe or prolonged neutropenia, those with a history of recent bone marrow transplantation, and those presenting with hypotension). Response rates for most of the other subgroups were also lower in the cefepime arm, except for patients with an underlying solid malignancy. M.O. comment: The M.O. notes that the original JCSR reviewers observed lower response rates for cefepime in patients with the highest risk of severe bacterial infection, including those with: (1) underlying hematologic malignancy, (2) severe or prolonged neutropenia, (3) history of recent bone marrow transplant, and (4) those presenting with hypotension. This information was reflected in the INDICATIONS AND USAGE section of the cefepime label.

Original language from the INDICATIONS AND USAGE section:

“Empiric Therapy for Febrile Neutropenic Patients. Cefepime is indicated for empiric monotherapy of febrile neutropenia. Antibiotic monotherapy may not be

appropriate in patients at high risk for severe infection (including patients with a history of recent bone marrow transplantation, with hypotension at presentation, with an underlying hematologic malignancy, or severe or prolonged neutropenia). Insufficient data exist to demonstrate the efficacy of cefepime monotherapy in such patients (See CLINICAL STUDIES).”

The current M.O. proposes that this information be more clearly presented in the label as follows.

Proposed draft language for the INDICATIONS AND USAGE section:

(b) (4)

The M.O. questions if the lower response rates for cefepime among patients with the highest risk for severe bacterial infection might have been due to the fact that cefepime has less anaerobic coverage than other agents generally used for empiric monotherapy in febrile neutropenia, namely, ceftazidime, piperacillin/tazobactam, and the carbapenems. (Such pathogens may have been more difficult to culture in prior studies.) Periods of severe stress, such as those associated with sustained hypotension and extreme immune suppression (s/p recent bone marrow transplant) might predispose susceptible patients to excessive translocation of gut anaerobic bacteria leading to bacteremia and sepsis. Therefore, the narrower spectrum of cefepime may be a double-edged sword in the case of empiric therapy. This may also explain why cefepime + metronidazole performed better than comparator regimens in trials for the treatment of complicated intra-abdominal infections.

The JCSR also noted the following. Microbiologic response for specific pathogens was similar between treatment arms in the pooled evaluable patient data set except for primary infections caused by Escherichia coli; there were more treatment failures for this organism in the cefepime arm. This difference was statistically significant under the primary analysis, which counted secondary infections unrelated to the initial episode as failures. The difference in response rates for E. coli infections was not statistically significant for other definitions of response. The most common reason for E. coli infection treatment failures included superinfection by a different, resistant organism; occurrence of fever without a microbiologically or clinically documented source of infection; and resistance of the original isolate. There were no cases in which there was persistence of a susceptible isolate. M.O. comment: The M.O. notes that the original JCSR reviewers observed that though there were more cefepime treatment failures caused by E. coli, these treatment failures were due to: (1) superinfection by a different, resistant organism; (2) occurrence of fever without a microbiologically or clinically documented source of infection; and (3) resistance of the original isolate. The JCSR reviewers noted that there were no cases of persistence of a susceptible isolate.

The JCSR also noted the following.

The trials comparing cefepime with combination therapy did not establish therapeutic equivalence between cefepime and the comparator arm. Both of these studies had relatively small numbers of evaluable patients and were therefore underpowered to demonstrate equivalence. In addition, one of these studies (AI411-118) had a point estimate of – [negative] 13.5% for the difference in response rates between cefepime and combination therapy; the size of this point estimate for this single study suggests that cefepime may have been inferior to the comparator regimen [piperacillin/gentamicin]. The studies comparing cefepime in combination with vancomycin or amikacin to the corresponding ceftazidime combination also did not establish therapeutic equivalence. Both of these studies, which were carried out in Europe, had small numbers of evaluable patients, due to study conduct involving use of prophylactic parenteral antibiotics in many study subjects, as well as failure to follow a substantial number of subjects beyond the end of therapy. Both studies showed lower response rates for the cefepime-containing treatment arm than for the comparator regimen. One study had a large point estimate for the difference in response rates between treatment arms, at -13.0%, while the other had a point estimate of -6.4%. Statistical analyses of these data suggest that cefepime combination therapy was inferior to the comparator regimen in these studies. Safety analysis: There were no significant differences between cefepime and comparator treatment arms with respect to overall mortality, mortality due to specific causes, clinical adverse events, or laboratory adverse events. M.O. comment: The M.O. noted that the original JCSR reviewers observed that there were no significant differences between cefepime and comparator study arms regarding overall mortality, mortality to specific causes, clinical adverse events, or laboratory adverse events. The JCSR concluded the following. Conclusions: The three studies comparing cefepime monotherapy to ceftazidime monotherapy were adequate and well-controlled. The two largest monotherapy trials, when pooled, and the third monotherapy trial, demonstrated therapeutic equivalence between cefepime and comparator for empiric therapy of febrile neutropenia, supporting the conclusion that cefepime monotherapy is safe and effective for this indication. The studies involving combination therapy with cefepime were not adequate to demonstrate efficacy of the combination regimen. Recommendations: 1. Approval of the claim of safety and effectiveness of cefepime monotherapy as empiric therapy for febrile neutropenia, with appropriate labeling to exclude use as monotherapy for patients at high risk for severe infection. 2. Non-approval of the claim of effectiveness of cefepime in combination as empiric therapy for febrile neutropenia. 3. A phase IV commitment to perform an adequate and well-controlled study examining the use of cefepime in combination with an aminoglycoside or vancomycin. M.O. comment: The JCSR reviewers explicitly recommended that cefepime labeling should exclude its use as monotherapy for patients at high risk for severe infection; and that cefepime combination therapy should not be approved as empiric therapy for febrile neutropenia.

The proposed Phase 4 commitment to further study cefepime in combination with an aminoglycoside or vancomycin was not in the approval letter for the indication of cefepime monotherapy as empiric therapy for febrile neutropenia. It is not clear to the M.O. why the Phase 4 commitment did not make it into the approval letter.

Regarding nonclinical pharmacology/toxicology, the JCSR stated the following. No new animal pharmacologic or toxicologic data were submitted with this supplement. The reader is referred to the animal pharmacology/toxicology review of the original NDA, by Dr. Kumar Mainigi et al., for a fuller summary of pharmacologic and toxicologic information. Acute toxicity studies In Sprague-Dawley rats, intravenous administration of cefepime causes a dose-dependent pressor response; at doses of 500 mg/kg or greater, this pressor response is followed by rapid systemic arterial collapse and death. High doses of cefepime also cause ataxia, respiratory distress, tremors, and seizures. LD50 values in rats are as follows: males, 1272 mg/kg; females, 1067 mg/kg. JCSR Medical Officer’s Comment The dosage schedule proposed in this supplement represents a dose of 28.6 mg/kg for a 70 kg individual, and 40 mg/kg for a 50 kg individual.

Subacute/chronic/carcinogenicity studies In a 26 week chronic toxicity study in dogs, cefepime caused thrombocytopenia, leukopenia, and severe anemia. Thrombocytopenia and leukopenia resolved by 2 weeks after discontinuation of drug; anemia resolved 7 weeks post treatment. Rechallenge with cefepime resulted in more rapid reappearance of hematologic abnormalities. No long-term carcinogenicity studies have been conducted on cefepime. Reproduction studies Cefepime is not embryocidal or teratogenic in rats at doses up to 1000 mg/kg/day, in mice at doses up to 1200 mg/kg/day, or in rabbits at doses up to 100 mg/kg/day. Mutagenicity studies Cefepime induces clastogenesis in primary lymphocytes cultured in vitro after 20 hours exposure, but not after 4 hours exposure.

M.O. comment: The results of the original nonclinical studies are consistent with the safety signals observed in the human experience.

Regarding clinical microbiology, the JCSR noted the following. No new microbiologic data were submitted with this supplement. The reader is referred to the microbiology review of the original NDA, by Dr. Pandu Soprey, for a fuller summary of microbiologic information. Cefepime is a semi-synthetic, broad spectrum cephalosporin antibiotic. It is sometimes referred to as a fourth- generation cephalosporin because of its antimicrobial spectrum, which is a function of its physicochemical and biochemical properties. Cefepime possesses an aminothiazole group, which enhances antipseudomonal activity. The presence of a zwitterionic methylpyrrolidino group results in high solubility in water and rapid penetration through bacterial cytoplasmic membranes. Like other cephalosporins, cefepime binds to and inactivates bacterial penicillin-binding proteins (PBPs), which are serine proteases required for cell wall synthesis; inactivation of PBPs leads to inhibition of cell wall synthesis, autolysis, and cell death. Cefepime is thus bactericidal. For most eubacteria, the PBP in lowest abundance is PBP2; cefepime has substantially higher affinity than do other cephalosporins for this protein in Escherichia coli and Enterobacter cloacae. Thus, saturation of PBP2 sites by cefepime occurs at a relatively low drug concentration. Cefepime binds poorly to PBP2 of Pseudomonas aeruginosa, but does have a high affinity for P. aeruginosa PBP3. In vitro, cefepime is active against most clinically relevant gram-negative aerobic bacteria, including most Enterobacteriaceae, ~90% of strains of P. aeruginosa, Haemophilus influenzae (both penicillinase-positive and negative), Moraxella catarrhalis, and Acinetobacter lwoffi. It is not active against non-aeruginosa species of Pseudomonas, and has only intermediate activity against Acinetobacter anitratus and Xanthomonas maltophilia. Cefepime is active against many pathogenic gram-positive aerobic bacteria, including methicillin-susceptible strains of Staphylococcus aureus, ~90% of strains of Staphylococcus epidermidis, Streptococcus pneumoniae (including penicillin-resistant isolates of S. pneumoniae), streptococci from Lancefield groups A, B, C, D (non-enterococcal), F, and G, and viridans streptococci. Like other cephalosporins, it is not active against methicillin-resistant S. aureus, Enterococcus species or Listeria monocyto-genes. With regard to relevant anaerobic bacteria, cefepime is active against Fusobacterium nucleatum, F. necrophorum, and Clostridium perfringens. It is not active against Clostridium difficile, and has intermediate or no activity against Bacteroides species, except for Bacteroides ureolyticus; Bacteroides fragilis, in particular, is resistant to cefepime. Cefepime has only intermediate activity against Peptostreptococcus species. JCSR Medical Officer’s Comment Cefepime is active against the majority of bacterial species likely to cause infection in neutropenic patients. However, it is not active against enterococci and methicillin-resistant S. aureus, which are potentially important sources of bacterial infection in this patient population. In some settings, use of cephalosporins has been associated with an increased incidence of enterococcal superinfection.

M.O. comment: The M.O. notes that cefepime lacks activity against B. fragilis.

Cefepime’s mechanism of action predicts that it should be bactericidal against susceptible species. This has been confirmed by time-kill analyses and determination of minimal bactericidal concentrations (MBCs). For gram-positive organisms, the MBC/MIC ratio is ~2, while for gram-negative organisms it ranges from 2.0 to 8.0. Bactericidal activity is less evident against methicillin-resistant or penicillinase-producing gram-positive staphylococci. Cefepime shows a dose-dependent post-antibiotic effect (PAE) against E. coli, E. cloacae, K. pneumoniae, S. marcescens, and S. aureus, but not against P. aeruginosa; the PAE, when present, appears to be 50-100% longer than that observed for ceftazidime.

Cefepime has been shown to be synergistic with aminoglycosides against some strains of gram-negative bacteria, but not consistently. Its action is antagonized by imipenem or polymyxin B. It does not affect opsonin-mediated phagocytosis of gram-negative bacteria, but in sub-MIC concentrations does increase the bactericidal effect of human serum against P. aeruginosa.

M.O. comment: The M.O. notes that the bactericidal action of cefepime is antagonized by imipenem and polymyxin B. The current Microbiology Team Leader, Frederic Marsik, Ph.D., noted the following in a discussion with the current M.O. Since imipenem targets the PBP like cefepime, it is possible that the two antibacterials might compete for the same sites of action. In the case of polymyxin it acts by destabilizing the cell wall and can actually insinuate itself into the wall so as with imipenem competition may exist that could interfere with the activity of the drug. However, the clinical significance of these potential interactions is unclear because there are no published cases of the use of cefepime with either drug documenting a failure of therapy.

The JCSR also noted the following: Cephalosporin resistance may arise due to bacterial synthesis of ß-lactamase; such synthesis may be constitutive or result from cephalosporin-induced derepression of ß-lactamase mRNA synthesis. Alternatively, mutations in the genes encoding PBPs can affect cephalosporin binding to these proteins, leading to resistance. Finally, decreased cell permeability to cephalosporins can lead to resistance.

Cefepime is not hydrolyzed by the major plasmid-mediated ß-lactamases, nor is it an effective inducer of ß-lactamase expression. Except for P. aeruginosa, cefepime is active against bacterial strains constitutively expressing a chromosomally-encoded ß-lactamase. Cefepime is also more likely to retain activity against bacteria which have acquired a plasmid-encoded ß-lactamase. Bacteria which are resistant to third generation cephalosporins such as ceftazidime generally remain susceptible to cefepime. Resistance to cefepime appears to require both decreased outer membrane permeability and high level cephalosporinase production. Selection for drug resistant isolates in vitro occurs at a lower rate for cefepime than for other cephalosporins. In a murine model of peritonitis, exposure of P. aeruginosa or E. cloacae to cefepime was much less likely to result in emergence of resistant isolates than was exposure to ceftriaxone.

Agar dilution, broth dilution, and macrotube broth dilution methods give similar results for assaying susceptibility to cefepime for a panel of reference strains. Susceptibility to cefepime has been defined as an MIC of <8 µg/mL, intermediate susceptibility as an MIC of 16 µg/mL, and resistance as an MIC of >32 µg/mL. Using a 30 µg cefepime disk, susceptibility testing by the Kirby-Bauer disk diffusion technique has been correlated with quantitative determination of MICs; breakpoints for disk diffusion susceptibility testing have been defined as a zone diameter of >18 mm for susceptibility, 15-17 mm for intermediate susceptibility, and <14 mm for resistance.

Cefepime has been demonstrated to have a protective effect in a number of animal models of infection. In particular, in mice with cyclophosphamide-induced neutropenia, cefepime was more active than ceftazidime or cefotaxime in the treatment of infections caused by Enterobacteriaceae, P. aeruginosa, or staphylococci.

M.O. comment: Dr. Frederic Marsik, the Microbiology Team Leader, requested that BMS provide updated breakpoint information on the following pathogens. Pseudomonas aeruginosa Escherichia coli Klebsiella pneumoniae Enterobacter cloacae Acinetobacter baumannii Acinetobacter species Serratia marcescens Enterobacter aerogenes Stenotrophomonas maltophilia Proteus mirabilis Klebsiella oxytoca Citrobacter freundii

M.O. comment: Dr. Frederic Marsik, Microbiology Team Leader, and Dr. Charles Bonapace, Human Biopharmaceutics Team Leader, in conjunction with an SGE, will be further evaluating the cefepime breakpoints for Pseudomonas aeruginosa and Enterobacteriaceae.

It is likely that their analyses will continue after the current review is completed.

Regarding human pharmacokinetics and pharmacodynamics, the JCSR noted the following. No new human pharmacokinetic or pharmacodynamic data were submitted with this supplement. The reader is referred to the biopharmaceutic review of the original NDA, by Dr. Ene Ette, for a fuller summary of biopharmaceutic information. Cefepime is rapidly absorbed and distributed following parenteral administration. Bioavailability following intramuscular injection is 100%. The drug is 16-19% protein-bound in plasma. Cefepime is metabolized to N methylpyrrolidine (NMP), which is converted to NMP-N-oxide. 88% of a dose of cefepime is excreted unchanged, 6.8% as NMP-N-oxide, 2.5% as the 7-epimer of cefepime, and ‹1% as NMP. These metabolites do not have significant anti-microbial activity. Excretion is primarily via renal filtration. Typical values for relevant pharmacokinetic parameters for cefepime are shown in Table 7.1. These were determined in normal volunteers after intravenous injection of a single dose. Table 7.1. Single dose pharmacokinetic data for cefepime2

Dose (mg) Cmax (µg/mL) t½ (h) AUCinf (µg•h/mL) Vss (L) 500 39.1 ± 3.5 2.00 ± 0.38 70.8 ± 6.7 16.6 ± 2.5 1000 81.7 ± 5.1 2.47 ± 0.73 148.5 ± 15.1 17.9 ± 1.8 2000 163.9 ± 25.3 1.99 ± 0.21 284.8 ± 30.6 17.3 ± 1.2

The pharmacokinetic studies submitted with the original NDA have shown that Cmax and AUCinf are dose-proportional. Special population studies have shown that after adjusting for creatinine clearance, age and gender do not affect cefepime pharmacokinetics. Peak plasma concentrations and the volume of distribution are not affected by renal insufficiency. There is a linear relationship between the clearance time for cefepime and creatinine clearance; t½ is approximately 2.2 h in patients with normal renal function and 13 h in patients with end-stage renal disease. The drug is removed by hemodialysis (HD), with a t½ of 2.3 h; a 3 h hemodialysis eliminates 68% of the drug. Continuous ambulatory peritoneal dialysis (CAPD) is less efficient at removing cefepime than HD (14.5 mL/min vs. 161 mL/min). Drug concentrations in dialysate remain above the MIC90 for clinically relevant organisms for at least 48 hours following a 2000 mg dose. JCSR Medical Officer’s Comment These pharmacokinetic data suggest that at a dosage schedule of 2 g IV q8h, serum concentrations of cefepime will remain at or above the MIC90 for the majority of bacterial species likely to infect neutropenic patients. Cefepime pharmacokinetics are not significantly altered in patients with serious bacterial infections or those with hepatic dysfunction. Co-administration of cefepime and amikacin does not change the pharmacokinetic profile of either drug.

M.O. comment: Dr. Charles Bonapace, the Human Biopharmaceutics Team Leader, requested that BMS provide cefepime pharmacokinetic parameters (e.g., plasma clearance, volume of distribution, elimination rate constant, and inter-compartmental constants) from healthy subjects from Phase 1 clinical trials using clinically-relevant doses and perform Monte Carlo simulations using requested MIC data (per Dr. Marsik) and pharmacokinetic data to determine the probability of target attainment with a PK/PD target of 60-70% T>MIC for each organism listed under MIC Data.

Dr. Frederic Marsik, Microbiology Team Leader, and Dr. Charles Bonapace, Human Biopharmaceutics Team Leader, in conjunction with an SGE, will be further evaluating the cefepime breakpoints for Pseudomonas aeruginosa and Enterobacteriaceae.

It is likely that their analyses will continue after the current review is completed.

Please refer to Appendix 7.3 for additional details related to the 1997 JCSR review.

3 DATA SOURCES, REVIEW STRATEGY, AND DATA INTEGRITY

3.1 Review Strategy

The analysis of mortality associated with cefepime versus comparable comparator agents was divided into several components.

1. The current M.O. • The current M.O. reviewed the case report forms of all febrile neutropenic patients who died within 30 days of receiving cefepime or a comparator agent in the nine febrile neutropenia studies (7 comparative and 2 non-comparative studies) submitted to the US FDA by BMS. The M.O. attempted to identify the most likely cause of death, as well as, major factors (co-morbidities, adverse events, documented pathogens) that may have contributed to each patient’s death. Based on the clinical and nonclinical safety profile of cefepime, adverse events of special interest were identified and reviewed. These included any AEs associated with death, particularly neurological impairment/seizure, renal toxicity, liver toxicity, study drug failure, and central nervous system hemorrhage. The M.O. tallied this information and evaluated it for potential mortality trends.

• As noted in Section 3.3 of this review, the current M.O. attempted to obtain information from BMS on the complete list of all clinical studies that included cefepime as a study agent. This encompassed all published and unpublished trials, as well as, all U.S. and non-U.S. trials including those not previously submitted to the FDA. Studies were characterized based on level of mortality data (patient vs. trial), whether mortality data were based on the Intent-to-Treat or the Clinically Evaluable population, whether mortality rates were based on actual patients vs. episodes of therapy (febrile neutropenia studies only), phase of study, study indication, comparator agent used (if applicable), combination regimen used (if applicable) use of blinding (if applicable), the country where the study was performed, duration of follow-up, and if the study was included in the Yahav et al. analysis. This process took over one year to accomplish, included numerous information requests, and culminated in BMS’s submission

of a revised “Master Table” of all cefepime clinical studies on 9/8/08. This information was eventually used, in part, to define the set of trials utilized by Yu-te Wu, Ph.D. from the Quantitative Safety and Pharmacoepidemiology group to perform her meta-analysis.

• In addition, the Agency was able to obtain mortality data from an additional, large (969 patient) Spanish study that was originally noted by Yahav et al.1 and later by BMS as having missing mortality data. This study by Sanz et al.17 compared cefepime versus piperacillin/tazobactam for empiric therapy for febrile neutropenia. Sumati Nambiar, M.D., M.P.H., Deputy Division Director of Safety for DAIOP, was able to contact a colleague of Dr. Sanz who provided the missing mortality data from this large, comparative Febrile Neutropenia study from Spain. The M.O. analyzed the mortality information provided from the co-author for potential mortality trends.

M.O. comment: The results of the Sanz et al. paper are discussed in further detail in the Section 5.1 of this review.

2. Collaborative Work with Ana Szarfman, M.D., Ph.D. • The M.O. sought the data mining expertise of Ana Szarfman, M.D., Ph.D., to aid with the following: (1) to further characterize the study population that encompassed the nine febrile neutropenia studies; (2) to search for any additional deaths, beyond those originally noted by BMS, that may have occurred among the patients enrolled in the 9 febrile neutropenia studies (this was necessary because CDISC standards do not have a specific variable for death and BMS’s calculations of deaths varied from one submission to the next); and (3) perform additional analyses of patient deaths, including evaluating for AEs associated with the deaths, associations between patient co morbid conditions and death, and the use of Bayesian meta-analytical methodology to assess potential causes of death (as developed by William DuMochel, Ph.D.).

• She performed a trial level analysis of 30-day mortality on the 41 studies already referenced by Yahav et al.1 plus an additional 47 studies. These 88 studies included 9467 cefepime-treated patients and 8288 comparator-treated patients.

[SEE THE SECOND ADDENDUM ATTACHED TO THIS REVIEW.]

• Dr. Wu performed a pre-specified analysis of patient level data on 35 studies. This third analysis included 5058 cefepime-treated patients and 3976 comparator- treated patients. Analyses of 30-day mortality were stratified based on the following variables: clinical condition (treatment indication), comparator drug, age, gender, race, location, any pathogen recovered at baseline, all pathogens isolated at baseline susceptible to study therapy, fungal pathogen recovered at baseline, baseline infection (mono- or polymicrobial), renal insufficiency or failure, active cancer or malignancy, and history of bone marrow transplant.

4. The Office of Surveillance and Epidemiology (OSE) (effort lead by Cynthia Kornegay, Ph.D.) was consulted and is in the process of designing a large case control study of mortality among febrile neutropenia patients on cefepime versus comparable agents (as noted in the Yahav paper) using Premier data (a hospital utilization and billing database). OSE is also providing ongoing comments to BMS on their proposed postmarketing analysis of mortality associated with the administration of cefepime versus comparable agents.

5. The relationship between cefepime’s pharmacokinetic properties and the current microbiology breakpoints is currently being reviewed by Frederic Marsik, Ph.D. (Clinical Microbiology Team Leader) and Charles Bonapace, Pharm.D. (Human Biopharmaceutics Team Leader). They requested that BMS provide: (1) recent MIC data for the specific pathogens, (2) pharmacokinetic parameters (e.g., plasma clearance, volume of distribution, elimination rate constant, and inter-compartmental constants) from healthy subjects from Phase 1 clinical trials using clinically-relevant doses, and (3) perform Monte Carlo simulations using the requested MIC and pharmacokinetic data to determine the probability of target attainment with a PK/PD target of 60-70% T>MIC for the specific pathogens. The pathogens of interest were: Pseudomonas aeruginosa, Escherichia coli, Klebsiella pneumoniae, Enterobacter cloacae, Acinetobacter baumannii, Acinetobacter species, Serratia marcescens, Enterobacter aerogenes, Stenotrophomonas maltophilia, Proteus mirabilis, Klebsiella oxytoca, and Citrobacter freundii. This review is ongoing.

3.2 The Yahav et al.1 paper

Yahav et al. evaluated 57 publications of comparative studies in the literature. The following flow chart provides the process used by the authors for identifying relevant publications.

In these studies, cefepime was compared with other β-lactams (comparative subgroups included: ceftazidime, piperacillin/tazobactam, imipenem or meropenem, and ceftriaxone or cefotaxime) in the treatment of febrile neutropenia, pneumonia, urinary tract/gynecologic infections, and other/mixed infections.

The following figure, derived from Yahav et al.1, provides the individual mortality results per study (publication) stratified on clinical condition for cefepime versus comparator, as well as, the overall result for the meta-analysis.

Figure 1. Yahav et al.’s Mortality Results Stratified on Study (Publication) and Clinical Condition for Cefepime vs. Comparator and Overall Result for the Meta-analysis (obtained from pg. 341 of Yahav et al.1)

The primary end-point was 30-day all-cause mortality post therapy. The authors found mortality data in 41/57 publications (Yahav et al. reported that 16 publications were excluded because of missing mortality information). They observed a higher 30-day all-cause mortality with cefepime as compared with the other β-lactams studied, risk ratio = 1.26 [95% CI 1.08, 1.49]. Based on clinical condition, the higher mortality was only observed in the febrile neutropenia studies. No significant difference in mortality was observed for the other clinical conditions.

M.O. comment: This finding was originally noted in a prior publication by the same group of authors (Paul et al., 2005)7 of a meta-analysis evaluating empiric antibiotic monotherapy for febrile neutropenia. Cefepime was associated with higher 30 day all-cause mortality than other β-lactams (imipenem, meropenem, ceftazidime, and piperacillin/tazobactam [RR 1.44, 95%CI 1.06-1.94, 3123 participants]. The following numbers of trials per comparator were evaluated: ceftazidime – 19 trials, cefepime – 18 trials, imipenem – 12 trials, meropenem – 8 trials, and piperacillin/tazobactam – 7 trials. Out of those evaluated, 28 studies had information on all-cause mortality. According to the authors, all-cause mortality was reported by intention-to-treat analysis in 13 of the 28 trials. The follow-up time for the mortality assessment was variable, ranging between the end of treatment to 30 days following the end of treatment. The authors commented that different lengths of follow-up may have introduced bias. The authors also noted that patients were included in the analysis more than once in 26 of 33 trials. This was due to recurrent episodes of febrile neutropenia in the same patient. The authors noted that this was methodologically incorrect since the statistical tests used assumed independence between observations. Regarding cefepime versus comparators, the authors found that bacterial superinfections tended to occur more frequently in patients on cefepime (RR 1.70, 95%CI 0.94, 3.09), and that more patients on cefepime discontinued treatment due to infections; however these findings did not reach statistical significance.

In the Yahav et al. paper, the authors found no significant differences between groups in treatment failure, superinfection, or adverse events.

The strengths of the Yahav et al. meta-analysis included: (1) use of a systematic approach for reviewing relevant publications, (2) the results of the meta-analysis were significant whether a fixed-effects model or random-effects model was used, (3) the test for publication bias was negative, and (4) a higher risk of mortality was observed when only studies with adequate allocation concealment and allocation sequence generation were included in the analysis.

The limitations of the Yahav et al. meta-analysis included: (1) lack of complete data on mortality, that is, mortality data were only available in 41/57 studies reviewed [7388/11723 (63%) of patients]. (2) The authors could not provide a biologically plausible explanation for the increased mortality noted in the analysis. They hypothesized that the mortality increase may have been due to unrecognized adverse events (e.g., undiagnosed status epilepticus), or possibly due to a discrepancy between in vitro and in vivo anti-bacterial activity. (3) The increased mortality in the febrile neutropenia clinical condition appeared to be driven by two studies: (a) a conference abstract by Gomez et al. (2001)8 and (b) an article in a peer-reviewed journal by Chandrasekar et al (2000).9 (4) The meta-analysis only compared cefepime versus other β-lactams; studies for clinical conditions without a β-lactam comparator, e.g.,

complicated intra-abdominal infections were not included. In fact, based on BMS analyses, for complicated intra-abdominal infection studies, 2% (11/598) of cefepime patients and 4% (18/429) of comparator patients died within 30 days post therapy. (5) Allocation concealment and allocation sequence generation were only described in 30/57 studies. (6) Most of the studies used in the meta-analysis were open-label (only 6 were double-blind and 8 were outcome-assessor blinded. (7) There is no comparable meta-analysis for a comparator drug. (8) Many of the studies used in the Yahav meta-analysis were also used in the Paul meta- analysis. It was notable that Paul et al. disclosed that employees from Pfizer and GSK helped by “…conducting independent searches for additional studies”, for the febrile neutropenia studies as noted.

M.O. comment: On 7/31/08, BMS reported to the FDA that 33 out of the 57 papers referenced by Yahav et al. were derived from 30 BMS-sponsored studies. Out of these 30 BMS studies, 17 were originally submitted to the FDA for registrational purposes. This included 9 of the 16 studies reported by Yahav et al. as having missing mortality data. Therefore, the following seven studies originally cited by Yahav et al. had missing mortality data: Chang, 199810 (severe infections; ceftazidime (CFTZ) comparator)-36 patients Chuang , 200211 (pediatric febrile neutropenia; CFTZ comparator)-40 patients Huang, 200212 (severe infections; CFTZ comparator)-? 52 patients Jiang, 200313 (moderate/severe lower resp tract infections; CFTZ comparator)-written in Chinese [No deaths were discussed in the publication per informal translation by Yan Wang, Ph.D.] Kieft, 199414 (serious bacterial infections/sepsis; (CFTZ comparator)-133 patients Lin, 200115 (adult pneumonia; (CFTZ comparator)-41 patients Sanz, 200217 (febrile neutropenia; cefepime+amikacin vs. CFTZ+amikacin)-969 patients.

FDA was eventually able to obtain mortality data for the largest study with missing data (Sanz et al.—discussed in further detail in Section 5.1 of this review). Additonally, BMS reported that they were not able to contact Gomez et al.8 for additional information related to their study. It is also unclear why Gomez et al. never published their research in a peer-reviewed journal and whether the results found in the abstract were final or interim study results.

The article by Chandrasekar et al (2000)9 was based on results from BMS-sponsored Study 204. Study 204 was reviewed as part of the original supplementary package for cefepime for febrile neutropenia and was considered pivotal for approval of cefepime’s febrile neutropenia indication by the clinical and statistical reviewers at the time for the original review. The current M.O. also reviewed the mortality data from Study 204 and concluded that the excess deaths in the cefepime arm (up to 30 days post therapy) were due to underlying leukemia, lymphoma, or other malignant neoplasms, and/ or co-morbid conditions, and not due to cefepime treatment failure or previously undetected adverse events.

Further analyses of trial level mortality data, including the results from: (1) BMS studies originally submitted to the U.S. FDA that may not have been submitted for publication and were therefore omitted from the Yahav analysis, as well as, (2) unpublished cefepime studies that were never submitted to the U.S. FDA are discussed in Dr. Yu-te Wu’s review (from the Quantitative Safety and Pharmacoepidemiology group). Highlights of Dr. Wu’s analysis are discussed in the Section 4.112 of this review.

3.3 Characterization of the BMS data repository for cefepime clinical trials

It took approximately one year for BMS to provide: (1) a complete listing of all cefepime clinical trials, and (2) the level of mortality data available per trial (patient level vs. trial level). This included all published and unpublished trials, as well as, all U.S. and non-U.S. trials including those not previously submitted to the FDA. The following chronicles the efforts to identify all of these trials and the level of data available from each of them.

On August 15, 2007, the Division requested the following information from BMS based on cefepime clinical trials: (1) mortality based on microbiologic data; (2) mortality by indication; (3) mortality stratified by monotherapy versus combination therapy; (4) mortality based on renal function; (5) mortality based on length of treatment; (6) timing and cause of death; (7) summary of mortality data from unpublished studies; (8) summary of data available on neurological adverse reactions associated with cefepime; (9) copies of all safety summaries previously submitted to NDA 50-679.

On November 7, 2007, the Division made the following information request. “In the meta-analysis published in Lancet Infectious Diseases, May 2007, mortality data were available from 41/57 studies (63% of the population). Does the Sponsor have mortality data on any of the 16 studies for which data were not available in the published literature?”

On 12/3/07, BMS responded and noted that 9 out of the 16 “missing studies” were sponsored by BMS and that information from these studies was submitted in the 10/26/07 submission. On 12/14/07, the Division requested the following additional information. “Regarding your 12/3/07 submission related to the mortality data on 9 out of the 16 missing studies, we were unable to find the mortality data for 3 of the 9 studies reported to have such mortality data. Please provide the mortality data (total number of patients in each treatment arm and the number of patients who died per treatment arm) for the following studies. (The Division notes that BMS references their response submitted to the FDA on 10/26/07, however we have been unable to find the data of interest.) CPM 23-93.002, Jehn et al. 1998 (febrile neutropenia) AI411-174, Saito et al. 1992 (pneumonia study) AI411-168, Saito et al. 1992 (chronic respiratory tract infections)”

BMS responded on 12/16/07 and provided explicit mortality data for the three studies in question.

M.O. comment: As of December 2007, mortality data were not available for the following 7 studies. Chang, 1998 (severe infections; ceftazidime comparator (CFTZ))-36 patients Chuang , 2002 (pediatric febrile neutropenia; CFTZ)-40 patients Huang, 2002 (severe infections; CFTZ)-?52 patients Jiang, 2003 (mod/severe lower resp tract infxns; CFTZ)-cannot find

Kieft, 1994 (serious bacterial infxns/sepsis; (CFTZ)-133 patients Lin, 2001 (adult pneumonia; (CFTZ)-41 patients Sanz, 2002 (febrile neutropenia; cefepime+amikacin vs. CFTZ+amikacin)-969 patients

On, January 23, 2008, the Division requested the following. “In a communication dated 12/3/07 from BMS to the Division, BMS noted that mortality data were not available for the following study of febrile neutropenic patients by Sanz et al (2002) in the Journal of Antimicrobial Chemotherapy. However, the Division notes on page 87 of the attached paper, under the "Acknowledgements" section, that: "This work was supported in part by a grant from Bristol-Myers Squibb, Princeton, NJ, USA…"

“Therefore, the Division requests that BMS obtain and provide the necessary mortality data (total number of patients in each treatment arm and the number of patients who died per treatment arm).”

On January 29, 2008, the Division requested the following. “Specify which studies in the published meta-analysis (1) are the studies for which BMS has mortality data. For example: The study by Kebudi, 2001 [ref 45 in the Yahav et al. (2007) paper] is derived from data from BMS study “AI411-…”. (1) Yahav et al. Efficacy and safety of cefepime: a systematic review and meta-analysis. Lancet Infect Dis. 2007 May;7(5):338-48.”

M.O. comment: BMS later revised their accounting of studies associated with papers cited by Yahav et al. In a submission dated 7/31/08, they noted that 33 out of the 57 papers cited by Yahav et al. were derived from 30 BMS studies, and 17 out of the 30 BMS studies were submitted to the FDA for registrational purposes.

On March 17, 2008, the Division made the following information request.

1. Regarding the 10-5-07 corrected BMS response, Table 1, entitled, “Comparative Cefepime Trials by Indication”, describe the BMS-sponsored comparative studies by indication as noted in “Table 1” in further detail based on the following by trial: a. Name of comparator therapy b. Form of study blinding: i. Double-blinded ii. Outcome assessor blinded iii. Open label c. If combination study therapy, provide the full regimen

2. Regarding the 10-5-07 corrected BMS response, Table 2, entitled, “Non- Comparative Cefepime Trials by Indication”, subsection, “Multiple Bacterial Infections”, summarize the following by trial: a. Further delineate the types of infections treated/trial. b. Were these trials for compassionate use? c. Provide the reasons for the seemingly elevated mortality rates.

3. Regarding the 10-26-07 BMS submission (2,272 pages), “Attachment Q1.2: List of Unpublished Studies” (pages 334-342), please reformat the table to conform with the following tables from the 10-5-07 corrected BMS response, Table 1, entitled, “Comparative Cefepime Trials by Indication”, and, Table 2, entitled, “Non-Comparative Cefepime Trials by Indication”, providing, by indication, study number, and treatment arm the following: a. total number of subjects enrolled b. number and percent of subjects who experienced all-cause mortality within 30 days of cefepime therapy c. totals for the above per indication d. Additionally, provide the following further detail by trial i. Name of comparator therapy ii. Form of study blinding: 1. Double-blinded 2. Outcome assessor blinded 3. Open label iii. Treatment indication iv. If combination study therapy, provide the full regimen v. Country or countries where the trial was (were) performed

4. Regarding the 3-7-08 BMS submission, “Table 1: Sponsorship of Cefepime Trials Analyzed in Published Meta-analysis”, please provide the following: a. A complete list of all 57 studies cited by Yahav et al. (41 are accounted for in the 3-7-08 table) i. We understand that BMS responded to the sponsorship issue for the 16 missing studies in the 12-3-07 submission, however for completeness, and to have all of the studies/publications accounted for in one table, please include in this revised table. b. In the table please stratify by treatment indication and include: i. Yahav et al. reference number ii. primary author and year published iii. whether or not BMS sponsored the study and BMS study number 1. indicate whether the study was submitted to the U.S. FDA for registration purposes c. Regarding the Yahav et al paper, Reference # 33, Gainer et al., 1995, Yahav et al. noted that the publication was missing mortality data, however, the 12-3-07 BMS submission on the 16 studies missing

mortality data from Yahav et al. did not account for this publication. Please delineate whether or not this study was sponsored by BMS and incorporate the pertinent information into your response to this question. d. Regarding the 12-3-07 BMS submission, under “Table 1: Status and Disposition of 16 Published References in the Cefepime Meta-analysis Publication”, “Reference #41 (Huang et al., 2002)”, according to Yahav et al., the authors were able to obtain mortality data for Huang et al. (2002), however, the authors were not able to obtain mortality data for Huang et al. (2005). Please clarify whether BMS has mortality data for Yahav et al. Reference #42 (Huang et al. 2005) and incorporate the pertinent information into your response to this question.

5. Provide a list of all available studies for which BMS has access to primary data (both published and unpublished, submitted to the FDA and not submitted to the FDA). In the list, delineate which studies were double- blinded, outcome assessor blinded, and open label studies.

6. Please clarify if our understanding of the extent of data BMS has access to as outlined below is accurate: Of the 57 publications included in the Yahav et al. meta-analysis, 25 publications were based on BMS sponsored studies. Additionally, there are 68 BMS sponsored studies (comparative and non- comparative) that were not submitted to US FDA for registrational purposes. Please identify by study number which of the 68 “Unpublished Studies” were submitted to non-U.S. regulatory authorities. For each study, identify the non-U.S. regulatory agency to which BMS submitted the study. Some of these studies may have been included in the meta-analysis.

BMS provided their responses to these questions on 7/31/08.

M.O. comment: BMS provided an initial response to most of the questions on 7/31/08. However, the Agency noted inconsistencies which led to a teleconference with BMS on 8/15/08. This led to a revision of the BMS response on 9/8/08 to incorporate corrections to the data, some of which are noted below. “Please note: Bristol-Myers Squibb has revised this response. The master table (Table 2) was revised to incorporate corrections to the data. Table 1 was also revised to reflect the addition/deletion of some studies. Finally, the text below was revised to reflect these changes.

Table 2 contains the requested information. The table is organized by those studies previously submitted to the FDA for registration (“BMS Clinical Trials”), those BMS-sponsored trials not submitted for significant regulatory review (“BMS Unpublished Studies”), and other BMS-sponsored studies that have been located (“Other BMS Studies”).

Unless otherwise noted, all deaths are within 30 days of study drug discontinuation.

Data in Table 2 were taken from the referenced reports, with the following exceptions: • One study in Table 2 (CPM 2296004) does not have mortality data; information is based upon a publication and no mortality data was included in it. • In 7 studies, the referenced report did not contain mortality information but it was instead obtained from the local BMS affiliate (AI411242, CPM6796008, CPM2293003, AI411174, CPM3694007, CPM0894005, and AI411172). • In 5 studies, no report was available and all of the data were provided by the local BMS affiliate (AI411197, AI411247/P01-401, AI411168, AI411165/166, and AI411177). • In 10 studies, no report was available and the only data BMS has is from the CARES database (AI411222, AI411226, AI411228, CPM0895010, CPM099601, CPM2394006, CPM2298004, CPM 0895009, CPM484, and CPMUT97). BMS has searched local archives and to date, has not been able to find other documentation for these studies.

The number of patients and deaths in Table 2 are based on the referenced reports and thus may not be the same as those cited in the meta-analysis, or in the CARES database or in previously submitted responses.

The initial list of 68 unpublished studies provided to FDA in the 26-Oct-07 response was reduced to 63 studies in the 13 May-08 response, and is being further reduced to 51 studies in this response. Table 1 outlines how the list of 68 studies was trimmed to 51.”

(b) (4)

On April 9, 2008, the Division and Sponsor held a teleconference. BMS agreed to: (1) attempt to contact the authors of the publications included in the meta-analysis that had missing mortality data; (2) provide further details about the 68 studies not submitted to the FDA (e.g., indications studied, mortality by treatment arm, and reasons for non-availability of data); (3) provide datasets for further analysis; and (4) conduct a meta-analysis that includes all publications that were included in the meta-analysis plus additional BMS sponsored studies that were not included in the meta-analysis.

On 5/13/08, BMS attempted to provide additional clarification on the 57 references in the meta analysis publication and the 68 non-registrational trials, including: (1) numbers enrolled per arm, (2) number of deaths per arm, (3) extent of available individual patient data, (4) explanation for why data were unavailable, e.g., author not responsive or BMS data not retained.

The Division responded with the following additional information requests on 5/21/08. 1. With reference to the 5/13/08 submission to the Agency concerning "Clarification on the 57 references in the (Yahav et al.) meta-analysis and the 68 non-registrational studies described by BMS", provide further clarification related to your responses to Questions 1 and 2.

For each of the studies described in your responses to Questions 1 and 2, clarify whether the total patient numbers and deaths for cefepime and

comparators were based on the Intent-to-Treat (ITT) population or the Clinically Evaluable population.

If the data presented for each of the studies were based on Clinically Evaluable populations, reformulate your responses based on the ITT populations. If ITT population data are not available for particular studies, provide explanations as to why the data are not available.

2. With reference to the 5/13/08 submission to the Agency provide further clarification related to your response to "Question 2":

On pages 5 and 6 of your response, you noted that 4 of the 68 "Unpublished" studies were originally listed twice and that one additional study, CPM089408, was incorrectly listed because it was not a study of cefepime, thus accounting for 63 "unpublished studies".

On page 7 of the response, you noted that 54 of the 63 studies were conducted at the regional or local level, and on page 8 you noted that 8 additional studies were "Global Clinical Studies". Please account for the 63rd study.

With reference to "Attachment Q2.1", on pages 15-24 of your response to Question 2, provide a timeline regarding when mortality information will be made available on the 11 studies for which BMS is awaiting further clarification on source data.

BMS provided their responses to the Division requests on 7/31/08.

M.O. comment: The submission on 7/31/08 was later revised by BMS and re-submitted on 9/8/08. Analysis of the submitted data is provided in the Safety section of this review.

On May 27, 2008, the Division made the following three requests. Request #1: By 7/31/08, the Agency wants a complete accounting of all studies that dealt with cefepime, i.e., a "Master Table" with a complete listing and tally of all available studies with (1) mortality level data and (2) patient level data. All studies should be counted only once whether they were included in the Yahav paper or not, whether originally submitted to the US FDA for registrational purposes or not. In your accounting, indicate which studies where referenced in the Yahav paper and which studies where submitted to the US FDA for registrational purposes. In addition, indicate whether data are based on the Intent-to-Treat (ITT) or Clinically Evaluable (CE) population.

If mortality level data are not available, provide an explanation as to why this information is not available. If patient level data are not available, provide an explanation as to why this information is not available. Please complete the following table.

Request #2: In order to ensure the internal quality of the patient-level data, we are asking that you perform a set of data checks. These checks are in addition to any quality procedures that you deem warranted. In the following, please refer to the April 30, 2008 request for patient-level data. If any discrepancies are revealed, ensure that the appropriate corrections are made prior to submitting the data to FDA. If corrections are not possible in certain cases, provide a summary of each discrepancy. The summary should include trial, and patient identification information using the variables STUDYID, and USUBJID and the same ID used in the patient-level dataset (PLD) and trial-level dataset (TLD) of April 30. Please refer to the following table on Quality Checks.

Request #3: Please reference the April 30, 2008 request for patient-level data, and the following BMS response sent in an e- mail to the Agency on 5/18/08:

"...Based on these discussions and Rho’s internal resource limitations, BMS will be able to deliver an integrated dataset (Tier I and Tier II) and trial-level data for the majority of the adult studies (26 of 31 studies) on 30-Jun 2008, with a cumulative integrated dataset containing the additional 5 adult studies delivered on 15-Jul-2008. An integrated dataset (Tier I and Tier II) for the 9 pediatric studies will be delivered on 31-Jul-2008."

Provide the integrated datasets and trial level data for all adult and pediatric studies in a single submission, please do not submit the information in a piece-meal manner as was proposed above. Provide a date for when this will be accomplished.

BMS provided their responses to the Division requests on 7/31/08.

M.O. comment: The “Master Table” of all the cefepime clinical studies submitted on 7/31/08 was later revised and resubmitted by BMS on 9/8/08. Analysis of the submitted data is provided in the Safety section of this review.

On August 15, 2008, the Agency and BMS held a teleconference. The Agency requested that BMS: (1) revise the “Master Table” submitted on 7/31/08 to only include mortality data up to 30 days post therapy and if it was not possible to discern < 30 day from > 30 day mortality for some studies then to indicate those studies in the “Master Table” to the Agency; (2) clarify when mortality numbers provided by BMS in the “Master Table” differed from those provided in the Yahav et al. meta-analysis; and (3) clarify for febrile neutropenia studies whether data were based on numbers of patients or numbers of febrile neutropenia episodes and when the number of patients was not available to indicate those studies in the “Master Table”.

M.O. comment: Several key issues were noted at this point in the review. (1) Prior to this submission, both the Yahav paper and BMS had provided mortality data based on the arbitrary cut-off point of 30 days post therapy, therefore, to maintain consistency, the Division requested that BMS continue to provide data based on 30-day mortality. (2) In a paper by Cherif et al.,16 the authors indicated that 10 deaths occurred in the study. However, the “Master Table” provided by BMS indicated only 7 deaths in the study. Further investigation revealed that 7 patients died due to infections (3 in the cefepime arm and 4 in the imipenem arm) and three additional patients died due to underlying leukemia or lymphoma progression. It was not clear whether the three additional patients were in the cefepime or imipenem arms. (3) Many of the febrile neutropenia studies were complicated by the fact that patients could be randomized more than once if they sustained additional episodes of febrile neutropenia. In some of these situations, the investigators counted episodes of febrile neutropenia instead of patients. This complicated calculating 30-day mortality when the denominator included episodes instead of patients.

On September 8, 2008, BMS submitted the revised “Master Table” of all clinical studies that contained a cefepime treatment arm.

The following Figure 2 is a flow diagram on which studies were selected for inclusion in the trial level analysis performed by Yu-te Wu, Ph.D.

Figure 2. Flow diagram for the trial level analysis

M.O. comment: For comparison, the Yahav analysis contained trial level mortality data on 3,931 cefepime patients and 3,457 comparator patients from 41 publications.

The following Figure 3 is a flow diagram on which studies were selected for inclusion in the patient level analysis performed by Yu-te Wu, Ph.D.

Figure 3. Flow diagram for the patient level analysis

M.O. comment: Dr. Wu’s patient level meta-analysis included 35 trials with 5,058 cefepime patients and 3,976 comparator patients. The Agency’s patient level analysis still included more patients than Yahav’s trial level analysis.

Please refer to Appendix 7.4 for additional details on the studies included in Dr. Wu’s analyses. In addition, the reader is referred to the BMS submission dated 9/8/08 (revised “Master Table”) for a complete list of all the cefepime trials provided by BMS.

4 INTEGRATED REVIEW OF SAFETY

4.1 Methods and Findings

The febrile neutropenia safety analysis set included all patients who received at least one dose of study drug in seven comparative studies (n= 1402 patients) and 2 non-comparative studies (n= 114 patients) for which BMS had patient level data. The comparative studies population consisted of 762 cefepime-treated patients and 640 comparator-treated patients.

The M.O reviewed the narratives and CRFs for all patients who died during the course of these studies. Based on this information, the M.O. created death narratives for all patients who died and attempted to determine the most likely cause(s) of death for each patient.

The M.O. sought the data mining expertise of Ana Szarfman, M.D., Ph.D., to aid with the following: (1) further characterization of the study population that encompassed the nine febrile neutropenia studies; (2) to search for any additional deaths, beyond those originally noted by BMS, that may have occurred among the patients enrolled in the 9 febrile neutropenia studies (this was necessary because CDISC standards do not have a specific variable for death and BMS’s numbers of deaths appeared to change from one submission to the next); and (3) perform additional analyses of patient deaths, including AEs associated with the deaths, associations between patient co-morbid conditions and death, and the use of Bayesian meta-analytical methodology to assess potential causes of death as developed by William DuMochel, Ph.D.

Based on the clinical and nonclinical safety profile of cefepime, adverse events of special interest were identified and reviewed. These included any AEs associated with death, particularly neurological impairment/seizure, renal toxicity, liver toxicity, study drug failure, and central nervous system hemorrhage.

M.O. comment: The Applicant initially coded adverse events using a variety of AE dictionaries. The M.O. asked the Applicant to recode the febrile neutropenia dataset using MedDRA terminology and utilize CDISC format for all major datasets to facilitate the use of WebSDM and other data mining tools. The Applicant complied with this request.

Demographics of the Febrile Neutropenia Patient Population

Studies 131, 189, 198, and 204 allowed patients with more than one episode of febrile neutropenia to be re-randomized to either cefepime or comparator treatment with subsequent episodes (of febrile neutropenia). Therefore, the following tables provide total patient numbers per study stratified by first versus last randomized treatment.

The following Table 1 provides the total patient numbers per study stratified by first randomized treatment.

Table 1. Nine Febrile Neutropenia Studies with Patient Level Data Supplied by BMS—patient numbers stratified by first randomized treatment (Safety Population).

Study Cefepime Comparator Total N % N % N ai411118 Cefepime vs Piperacillin + Gentamicin 1:1 59 50.86 57 49.14 116 ai411131 Cefepime vs Ceftazidime 1:1 94 48.45 100 51.55 194 ai411137 Cefepime vs Mezlocillin + Gentamicin 1:1 35 49.3 36 50.7 71 ai411143 Uncontrolled 84 100 0 84 ai411158 Uncontrolled 30 100 0 30 ai411186 Cefepime + Amikacin vs Ceftazidime + Amikacin 2:1 242 68.56 111 31.44 353 ai411189 Cefepime vs Ceftazidime 1:1 139 49.47 142 50.53 281 ai411198 Cefepime + Vancomycin vs Ceftazidime + Vancomycin 1:1 53 47.75 58 52.25 111 ai411204 Cefepime vs. Ceftazidime 1:1 137 49.64 139 50.36 276 Total only Comparative Studies 759 54.14 643 45.86 1402 Total All Studies 873 57.59 643 42.41 1516

M.O. comment: Of note, Studies AI411-143 and 158 were uncontrolled. Study 186 had a 2:1 (cefepime+amikacin to ceftazidime+amikacin) randomization scheme. The Studies with the highest enrollment were: 186, 204, and 189.

The following Table 2 provides the total patient numbers per study stratified by last randomized treatment.

Table 2. Nine Febrile Neutropenia Studies with Patient Level Data Supplied by BMS—patient numbers stratified by last randomized treatment (Safety Population).

Study Cefepime Comparator Total % of % of N total N total N ai411118 Cefepime vs Piperacillin + Gentamicin 1:1 59 50.86 57 49.14 116 ai411131 Cefepime vs Ceftazidime 1:1 104 53.61 90 46.39 194 ai411137 Cefepime vs Mezlocillin + Gentamicin 1:1 35 49.3 36 50.7 71 ai411143 Uncontrolled 84 100 0 84 ai411158 Uncontrolled 30 100 0 30 ai411186 Cefepime + Amikacin vs Ceftazidime + Amikacin 2:1 242 68.56 111 31.44 353 ai411189 Cefepime vs Ceftazidime 1:1 144 51.25 137 48.75 281 ai411198 Cefepime + Vancomycin vs Ceftazidime + Vancomycin 1:1 54 48.65 57 51.35 111 ai411204 Cefepime vs Ceftazidime 1:1 138 50.00 138 50.00 276 Total only Comparative Studies 776 55.14 626 44.86 1402 Total All Studies 890 58.51 626 41.49 1516

M.O. comment: As compared with the analysis based on treatment assignment during the first episode of febrile neutropenia (first randomized treatment), the following table shows the patient shifts noted during the last randomized treatment in Studies 131, 189, 198, and 204.

Table 3. Changes in Treatment Allocation Between First and Last Randomization

Number of patients that Patient balance vs. analysis Study changed treatment allocation based on 1st randomization ai411131 Cefepime vs Ceftazidime 1:1 24 + 10 for cefepime ai411189 Cefepime vs Ceftazidime 1:1 19 + 5 for cefepime ai411198 Cefepime + Vancomycin vs Ceftazidime + Vancomycin 1:1 9 + 1 for cefepime ai411204 Cefepime vs Ceftazidime 1:1 15 + 5 for ceftazidime

Total only Comparative Studies 67 + 11 for cefepime

Total All Studies 67 + 11 for cefepime

M.O. comment: Therefore, after the final randomized treatment is taken into account, a net addition of 11 patients moved into the cefepime arm. Three of the 67 patients who changed treatment assignment between the first and last episode of febrile neutropenia died. These three patients (411-204-005-393 re-enrolled as 411-204-005-471, 411-204-010132 re-enrolled as 411-204-010-135, and 411-204-016-401 re-enrolled as 411-204-016-484) were in Study 204, and were initially enrolled as cefepime patients during their first episode of febrile neutropenia; however were later re-enrolled in the ceftazidime arm during a subsequent episode and died within 30 days of receiving ceftazidime.

Among the seven comparative febrile neutropenia studies, the following figures provide graphic presentations of age in years, sex, race, and country of origin based on the first randomized treatment (for ease of reference, the color coded key both precedes and follows the figures).

Figure 4. Age (in years) by Treatment Arm

Figure 5. Sex by Treatment Arm

Figure 6. Race by Treatment Arm

Figure 7. Country by Treatment Arm

M. O. comment: In the graphic representations of sex, race, and country, the x-axis = the total number of patients. Patients in the comparative studies appeared to be balanced with regard to sex, age, and race. There appeared to be more cefepime patients versus comparator patients from France. This was likely due to the fact that Study 186 was performed in France and had a 2:1 (cefepime/amikacin: ceftazidime/amikacin) randomization.

4.1.1 DEATHS

This M.O. reviewed the CRFs of all patients who BMS originally noted as having died in the nine febrile neutropenia studies. The M.O. attempted to determine if any association could be determined on retrospective review of the CRFs. The review included evaluation of the Applicant’s original Death Narratives and case report forms, as well as evaluation of case patients using Patient Profiles and WebSDM with the aid of Dr. Ana Szarfman. The M.O. wrote a new Death Narrative for each patient who died within 30 days of the last dose of study drug and provided comment on the most likely cause and contributing factors for each patient’s death.

The M.O. also attempted to identify if there were any additional cefepime or comparator arm patients who died within 30 days of the last dose of study drug, for whom BMS did not supply a Death Narrative. This was done in WebSDM with the help of Dr. Ana Szarfman by cross checking all study patients against dataset variables that indicated whether a patient (1) had an adverse event that resulted in death, or (2) had values="yes" or "no" for the disposition variable "death related to the administration of study drug". Using this process, a total of 141 unique subject identifying numbers (USUBJIDs) were identified for patients who died during the course of the studies. Out of these 141 USUBJIDs, there were 11 patients who were assigned two different USUBJIDs because they were enrolled twice for two separate episodes of febrile neutropenia. This left 130 patients who died, each with one unique subject identifying number. Out of the 130 remaining patients, 16 patients died more than 30 days after the last dose of study therapy. This left 114 patients who died within 30 days of the last dose of study therapy and who had a single unique subject identifying number.

The following Table 4 displays the 130 patients who died during the course of study therapy and follow-up. Rows highlighted in yellow identify patients who died more than 30 days post last dose of assigned study therapy. Rows highlighted in pink identify patients who died within 30 days of the last dose of study therapy and who were assigned more than one unique subject identifying number (USUBJID).

Table 4. One-Hundred Thirty Patients who Died during the Course of Study Therapy and Follow-up in the 9 Febrile Neutropenia Studies

missing from original BMS list of deaths or enrolled Died under >30 No multiple days No death Date of USUBJID Site ID Sex Age Race Initial Planned Arm USUBJIDs p tx Comment CRF narrative Death (b) (6) 411118001007 001 M 76 W Cefepime x x died 39 days p tx x

411118001011 001 F 57 W Gentamicin/Piperacillin

411118001020 001 M 37 W Cefepime

411118001022 001 M 68 W Cefepime

411118001038 001 M 63 H Gentamicin/Piperacillin

411118001047 001 M 22 B Cefepime

411118001065 001 F 24 W Cefepime

411118002017 002 M 72 W Cefepime

411118002029 002 F 57 W Gentamicin/Piperacillin

411118002047 002 F 46 W Gentamicin/Piperacillin

Shands/University 411131001009 of Florida F 68 B Cefepime

Shands/University 411131001013 of Florida F 55 W Cefepime

Shands/University 411131001016 of Florida F 62 B Ceftazidime

Shands/University 411131001033 of Florida M 43 W Cefepime

(b) (6) Shands/University 411131001034 of Florida F 25 B Ceftazidime

died > 30 days after last dose ceftaz with sepsis, severe Shands/University pancytopenia, 411131001041 of Florida M 43 W Ceftazidime x x refractory ALL. x x

Shands/University 411131001055 of Florida M 29 W Cefepime

Shands/University 411131001060 of Florida F 58 W Cefepime

Shands/University 411131001066 of Florida M 62 W Ceftazidime

Shands/University 411131001072 of Florida M 72 W Ceftazidime

Shands/University 411131001075 of Florida M 31 W Cefepime

also enrolled Shands/University as 411131001076 of Florida M 33 W Ceftazidime 411131001089

also enrolled Children's as 411131003008 Medical Center M 2 H Ceftazidime 411131003020

Children's 411131003016 Medical Center F 7 H Cefepime

also enrolled (b) (6) Children's as 411131003019 Medical Center M 10 O Cefepime 411131003057

Children's died 125 days post 411131003043 Medical Center M 12 H Ceftazidime ceftazidime

Children's 411131003053 Medical Center M 1 W Cefepime died 46 days post 411137001008 001 F 27 W Cefepime x x cefepime x

411137001031 001 F 24 W Mezlocillin/Gentamicin died 31 days post 411143001011 001 M 60 W Cefepime x x cefepime x also enrolled as died 39 days post 411143001013 001 M 21 W Cefepime 411143001036 x cefepime

411143001029 001 M 63 W Cefepime also enrolled as 411143001059 001 M 73 W Cefepime 411143001068

411143002005 002 M 60 W Cefepime

411143002009 002 M 58 O Cefepime also enrolled as 411143002015 002 M 18 O Cefepime 411143002021

411143002017 002 F 84 W Cefepime

411143002038 002 F 44 W Cefepime

411143002040 002 F 58 W Cefepime

(b) (6) 411158001001 001 M 45 W Cefepime

411158001011 001 F 42 W Cefepime

411158001017 001 F 44 W Cefepime

411158001025 001 F 22 W Cefepime

411158001027 001 M 53 W Cefepime x died 54 days post tx

411186002006 002 F 45 Cefepime

411186002007 002 F 62 Cefepime

411186004006 004 F 56 Cefepime

411186004007 004 M 42 Cefepime

411186004019 004 M 57 Ceftazidime

411186004021 004 M 42 Cefepime

died >30 days post cefepime with coma & 411186008009 008 F 46 Cefepime x x acute leukemia x x

411186010005 010 F 52 Cefepime x x died 33 days post tx x

411186011003 011 M 69 Ceftazidime died > 30 days post 411186011005 011 M 79 Cefepime x x cefepime x x died > 30 days post 411186012003 012 F 42 Ceftazidime x ceftazidime x x

411186015004 015 M 58 Cefepime

411186020004 020 F 45 Cefepime

411186020011 020 M 61 Cefepime

(b) (6) 411186023001 023 M 36 Ceftazidime

411186026009 026 F 39 Cefepime

411186026011 026 F 37 Cefepime

411186027007 027 F 30 Ceftazidime

411186028001 028 F 38 Cefepime

411186028012 028 M 40 Cefepime

411186029002 029 M 28 Cefepime

411186030001 030 M 47 Ceftazidime x x died 37 days post tx x

411186036003 036 F 33 Cefepime x x died 34 days post tx x

died >30 days post cefepime, Addendum: per BMS died on (b) (6) > 30 days post cefepime. unknown date of death but appeared to be alive at least to 7/30/93(last dose cefepime was 7/8/93), 411186036005 036 F 32 Cefepime x x died of viral infection x

411189003002 003 M 58 W Cefepime

411189005025 005 M 31 W Cefepime

411189006014 006 M 48 B Ceftazidime died > 30 days post 411189006023 006 F 50 W Cefepime x x cefepime x

also enrolled as (b) (6) 411189007010 007 M 37 W Ceftazidime 411189007015

411189009012 009 M 69 W Cefepime

411189009041 009 F 56 W Ceftazidime

411189009044 009 F 31 W Ceftazidime

411189010026 010 M 60 W Ceftazidime

411189010027 010 M 43 W Ceftazidime

411189010032 010 M 46 W Cefepime

411189016005 016 M 32 W Ceftazidime

411189016011 016 F 65 W Ceftazidime

411189018010 018 M 74 W Cefepime

411189019011 019 M 62 W Cefepime

411189022001 022 F 68 W Cefepime

411189022003 022 M 65 W Ceftazidime

411189022013 022 M 73 W Cefepime

411189023001 023 M 64 W Ceftazidime Ceftazidime Plus 411198001003 001 M 75 W Vancomycin Ceftazidime Plus 411198001014 001 M 60 W Vancomycin Ceftazidime Plus 411198002005 002 F 61 W Vancomycin also enrolled Cefepime Plus as 411198002012 002 M 50 W Vancomycin 411198002018

Cefepime Plus (b) (6) 411198002026 002 F 68 W Vancomycin Cefepime Plus 411198002029 002 F 26 W Vancomycin Cefepime Plus 411198004013 004 F 68 W Vancomycin Cefepime Plus 411198004029 004 M 49 W Vancomycin

411204003042 003 F 65 W Cefepime

411204003341 003 M 57 W Cefepime

411204005142 005 F 56 W Cefepime

411204005335 005 M 63 B Cefepime

BMS confirmed that pt was re-randomized as 411204005471& received 1 dose ceftaz prior to death on also enrolled (b) (6) . Initially as received cefepime 9/2- 411204005393 005 F 37 W Cefepime 411204005471 9/13/93. x

411204005395 005 F 61 W Ceftazidime

411204005472 005 F 41 W Cefepime

411204005475 005 F 48 O Cefepime

411204005491 005 M 54 W Ceftazidime

411204005559 005 M 62 B Cefepime

411204007022 007 F 23 W Cefepime

According to 8 15 08 BMS Unclear when pt died. response, Last dose cefepime on this patient 4/27/94, sepsis started died>30days 411204008352 008 M 61 W Cefepime x x 4/27/94. post tx (b) (6) 411204010067 010 M 78 W Ceftazidime

411204010070 010 F 45 W Cefepime

According to the “CRF Discrepancy/Resolution Form” at the end of the CRF, this patient #132 was the same as patient #135. In the protocol, a patient was apparently allowed to re-enroll into Study 204 seven days after completion of prior study therapy. The patient re-entered the also enrolled study on ceftazidime 18 as days post cefepime 411204010132 010 M 73 W Cefepime 411204010135 therapy.

411204010133 010 M 65 W Cefepime

411204010186 010 F 51 W Cefepime

411204010188 010 F 50 W Ceftazidime

411204010189 010 F 81 B Ceftazidime

411204010366 010 F 46 W Cefepime

(b) (6) 411204010369 010 M 53 W Ceftazidime

411204010561 010 M 72 W Cefepime

411204012373 012 F 57 W Cefepime

411204013161 013 F 62 B Ceftazidime

411204014117 014 M 74 W Cefepime

411204016102 016 F 82 W Cefepime

Addendum: BMS confirmed that patient initially received cefepime 9/4-9/10/93 and then was re- randomized as 411204016484 & also enrolled received ceftazidime as from 2/13-2/16/94. Died 411204016401 016 F 65 W Cefepime 411204016484 (b) (6)

411204016402 016 M 43 W Cefepime

411204016406 016 M 61 W Ceftazidime

411204016455 016 M 37 W Cefepime

411204016485 016 M 31 B Ceftazidime

411204016490 016 F 18 W Cefepime

411204016541 016 M 45 W Cefepime

411204017424 017 F 27 B Ceftazidime

411204018233 018 F 48 W Cefepime

In the seven comparative febrile neutropenia studies, there were 102/1402 (7.3%) deaths within 30 days of study therapy, regardless of treatment assignment. Based on treatment assignment from the first episode of febrile neutropenia, out of these 102 deaths, 8.4% (64/765) were cefepime patients and 6.0% (38/637) were comparator patients. Based on treatment assignment from the last episode of febrile neutropenia, out of these 102 deaths, 7.9% (61/776) were cefepime patients and 6.5% (41/626) were comparator patients. In addition, there were 12 patients who died in the two non-comparative studies.

The following Table 5 displays the febrile neutropenia study population stratified by treatment arm and death.

Table 5. Febrile Neutropenia Study Population by Treatment and 30-Day Mortality Status (based on study drug assignment during last episode of febrile neutropenia prior to death*) Study Deaths Total No. of Patients/Study Cefepime Comparator Total Cefepime Comparator Total n % (n/N) n % (n/N) n N N N ai411118 Cefepime vs Piperacillin + Gentamicin 1:1 5 8.47 4 7.02 9 59 57 116 ai411131 Cefepime vs Ceftazidime 1:1 9 8.65 6 6.67 15 104 90 194 ai411137 Cefepime vs Mezlocillin + Gentamicin 1:1 0 0.00 1 2.78 1 35 36 71 ai411143 Uncontrolled 8 9.52 0 8 84 0 84 ai411158 Uncontrolled 4 13.33 0 4 30 0 30 ai411186 Cefepime + Amikacin vs Ceftazidime + Amikacin 2:1 13 5.37 4 3.60 17 242 111 353 ai411189 Cefepime vs Ceftazidime 1:1 8 5.56 10 7.30 18 144 137 281 ai411198 Cefepime + Vancomycin vs Ceftazidime + Vancomycin 1:1 5 9.26 3 5.26 8 54 57 111 ai411204 Cefepime vs Ceftazidime 1:1 21 15.22 13 9.42 34 138 138 276 Total only Comparative Studies 61 7.86 41 6.54 102 776 626 1402 Total All Studies 73 8.20 41 6.55 114 890 626 1516

M.O. comment: The comparative study with the highest percentage of deaths was Study 204 (15.6% for cefepime and 9% for ceftazidime). Further analysis of the CRFs and clinical trial data of all the patients who died in the febrile neutropenia studies follows.

The following Tables 6 and 7 summarize the M.O.’s assessment of the most likely etiologies for patient deaths in the comparative and non-comparative studies, respectively.

Table 6. M.O. Categorization of Febrile Neutropenia Patient Deaths < 30 Days Post Therapy, Based on Last Study Therapy Prescribed prior to Death [Comparative Studies*] CEFEPIME n/61 n/776 n/41 n/626 a COMPARATOR Likely Cause of Death (N=61 ) a,b (N=41 ) TOTAL CODE Underlying malignancy/co-morbid condition 26 42.6% 3.4% 20 48.8% 3.2% 46 1 Possible study drug failure 3 4.9% 0.4% 3 7.3% 0.5% 6 2 Cannot rule out study drug failure, though 12 19.7% 1.5% 7 17.1% 1.1% 19 11 additional factors may have contributed to the patient's death. Death by resistant pathogen Gram Positive 5 8.2% 0.6% 2 4.9% 0.3% 7 4A Fungal Death 7 11.5% 0.9% 6 14.6% 1.0% 13 5 CNS bleed 5 8.2% 0.6% 1 2.4% 0.2% 6 6 Other infectious etiology Viral 2 3.3% 0.3% 2 4.9% 0.3% 4 7A Late bacterial 1 1.6% 0.1% 0 0.0% 0.0% 1 7B TOTAL 61 100.0% 7.9% 41 100.0% 6.5% 102

*There were 1402 patients in the seven comparative studies. Based on last randomized treatment, 776 received cefepime & 626 received

comparator. Out of these patients, 102 died within 30 days post therapy (61 cefepime & 41 comparator patients). a

Patient AI411-204-010-135 [a.k.a. AI411-204-010-132] was initially assigned by BMS as a cefepime death; however, the patient had a

second episode of febrile neutropenia and was enrolled in the ceftazidime arm prior to death. b

Paitent AI411-204-005-471 [a.k.a. AI411-204-005-393] and Patient AI411-204-016-484 [a.k.a. AI411-204-016-401] were initially assigned

by BMS as cefepime survivors; however, the patients each had a second episode of febrile neutropenia and were enrolled in the

ceftazidime arm prior to death.

M.O. comment: The difference in the proportion of deaths between cefepime and comparators in the 7 comparative studies was minimal, i.e., 1.4%. The majority of patients in both groups appeared to die due to underlying malignancy and/or co-morbid conditions. Among those that died, up to 24.6% of cefepime patients and 24.4% of comparator patients may have potentially died due to study drug failure. Slightly more cefepime patients appeared to die due to a resistant Gram-positive pathogen (8.2% of the cefepime deaths) as compared with the comparator patients (4.9% of the comparator deaths). Additionally, 5/61 of the deaths and 1/41 of the comparator deaths appeared to be due to a CNS bleed. This may have been a marker of severity of underlying illness and/or chemotherapy-induced thrombocytopenia.

Table 7. M.O. Categorization of Febrile Neutropenia Patient Deaths < 30 Days Post Therapy [Non-Comparative Studies*] Likely Cause of Death CEFEPIME (N=12) n/12 n/114 CODE Underlying malignancy/co-morbid condition 7 58.3% 6.1% 1 Possible study drug failure 1 8.3% 0.9% 2 Cannot rule out study drug failure, though 2 16.7% 1.8% 11 additional factors may have contributed to the patient's death. Death by resistant pathogen Gram Positive 0 0.0% 0.0% 4A Fungal Death 2 16.7% 1.8% 5 CNS bleed 0 0.0% 0.0% 6 Other infectious etiology Viral 0 0.0% 0.0% 7A Late bacterial 0 0.0% 0.0% 7B TOTAL 12 100.0% 10.5% *There were 114 patients in the two non-comparative studies.

M.O. comment: The most common cause of death in the non-comparative studies was underlying malignancy and/or co-morbid condition. One patient possibly died due to cefepime failure. In addition, the M.O. could not rule out cefepime drug failure in an additional two patients however, additional factors may also have contributed to the patients’ deaths, including metastatic cancer and disseminated fungal infection.

The following Table 8 provides additional details on the M.O.’s assessment of the likely etiologies and contributing factors associated with patient deaths (based on last randomized treatment prior to death). Please note that the code in the last column of the following table corresponds to the same coding used in the prior two tables to identify how the M.O. categorized each patient’s most likely cause of death. For emphasis, rows highlighted in yellow correspond to code 2 (possible study drug failure) and rows highlighted in pink correspond to code 11 (cannot rule out study drug failure, though additional factors may have contributed to the patient’s death).

The discussion of the death narratives follows thereafter.

Table 8. M.O. Likely Etiologies and Contributing Factors for Patient Deaths < 30 Days Post Study Therapy in Febrile Neutropenia Studies

Study AI411-118 CEFEPIME (CFP) TREATMENT ARM Patient Likely Cause of Death Major Contributing Factor(s) CODE ID AI411- Cardiac arrest. Died 19 days post Hypoxia secondary to radiation 1 118-001- therapy (CFP 2gIVq8h). pneumonitis, refractory Hodgkin's 020 lymphoma. AI411- Myocardial infarction. Died 2 days Intracerebral hemorrhage secondary to 1 118-001- post therapy (CFP 2gIVq8h). thrombocytopenia likely related to 022 leukemia & subsequent chemotherapy. AI411- Fungal sepsis (C. albicans). Died 21 Multiple organ failure, refractory leukemia. 5 118-001- days post therapy (CFP 2gIVq8h). 047 AI411- P. aeruginosa septic shock, Cardiac Cannot rule out study therapy failure; 11 118-001- arrest. Died within hours of 1st however, 4 hour delay from enrollment to 065 dose of CFP 2gIVq8h. 1st dose of cefepime. Widely metastatic neuroectodermal pituitary brain tumor. AI411- Cardiopulmonary arrest. Died 15 Adenocarcinoma of stomach, s/p 1 118-002- days post therapy (CFP 2gIVq8h). hemigastrectomy 14 days prior to death. 017 Possible intra-abdominal infection or pneumonia post cefepime.

Table 8. M.O. Likely Etiologies and Contributing Factors for Patient Deaths < 30 Days Post Study Therapy in Febrile Neutropenia Studies (cont'd)

Study AI411-118 COMPARATOR TREATMENT ARM [piperacillin & gentamicin (PIP/GENT)] Patient Likely Cause of Death per M.O. Major Contributing Factor(s) per M.O. CODE ID AI411- Septic shock. Died 6 days post PIP & Multilobar pneumonia, GI lymphoma with 11 118-001- 9 days post GENT. pancreatic mass, coagulase-negative 011 Staphylococcal bacteremia. Cannot rule out study drug failure, though additional factors may have contributed to the patient's death.

AI411- Multi-organ failure. Died 4 days post Lymphoma in an accelerated phase. 1 118-001- PIP/GENT. Cannot rule out that study therapy may 038 have contributed to renal & hepatic failure.

AI411- Hemorrhagic gastritis & CHF. Died 18 Acute leukemia, pancytopenia. Pulmonary 1 118-002- days post PIP/GENT. granulomas vs. fungal pneumonia. 029 AI411- Polymicrobial sepsis (Candidemia & Acute leukemia. 5 118-002- S. maltophilia). Died 18 days post 047 PIP/GENT.

Table 8. M.O. Likely Etiologies and Contributing Factors for Patient Deaths < 30 Days Post Study Therapy in Febrile Neutropenia Studies (cont'd)

Study AI411-131 CEFEPIME (CFP) TREATMENT ARM Patient Likely Cause of Death Major Contributing Factor(s) CODE ID AI411- Metastatic breast and uterine cancer. 1 131-001- Died 19 days post therapy (CFP 009 2gIVq8h). AI411- Sepsis (pathogen unknown), possible AML, CNS bleed. Cannot rule out study 11 131-001- secondary pneumonia with multi- drug failure, though additional factors 013 organ failure. Died 12 days post may have contributed to the patient's therapy (CFP 2gIVq8h). death. AI411- Pulmonary/disseminated 5 131-001- aspergillosis. Died 11 days post 033 therapy (CFP 2gIVq8h). AI411- Relapsed CNS leukemia. Died 9 days 1 131-001- post therapy (CFP 2gIVq8h). 055 AI411- Cardiopulmonary arrest. Died on Septic shock with acute abdomen, 11 131-001- therapy day 6 (CFP 2gIVq8h). lymphoma. Cannot rule out study drug 060 failure, though additional factors may have contributed to the patient's death. AI411- Pulmonary hemorrhage and Bone marrow failure, multi-organ failure, 1 131-001- respiratory failure. Died 19 days post Ewing's sarcoma. 075 therapy (CFP 2g IVq8h). AI411- Pediatric. Intracranial hemorrhage. Chemotherapy-induced 6 131-003- Died 1 day post therapy (CFP thrombocytopenia, metastatic 016 50mg/kgIVq8h). neuroblastoma. AI411- Pediatric. ARDS/RSV pneumonia. AML, C. albicans colitis, P. 7A 131-003- Died 11 days post therapy (CFP aeruginosa/enterococcal/coagulase- 053 50mg/kgIVq8h). negative Staph. bacteremia AI411- Pediatric. Stomatococcus Acute leukemia, profoundly neutropenic. 4A 131-003- mucilaginosus [Also enrolled as 411-131-003-019, 057 sepsis/Cardiopulmonary arrest. Died received Cefepime] 5 days post therapy (CFP 50mg/kgIVq8h).

Table 8. M.O. Likely Etiologies and Contributing Factors for Patient Deaths < 30 Days Post Study Therapy in Febrile Neutropenia Studies (cont'd)

Study AI411-131

COMPARATOR TREATMENT ARM [ceftazidime (CFTAZ)] Patient Likely Cause of Death per M.O. Major Contributing Factor(s) per M.O. CODE ID AI411- Subdural hematoma. Died on Thrombocytopenia, CLL. Coma with 6 131-001- therapy day 28 (CFTAZ 2gIVq8h). seizures. Cannot rule out that worsening 016 thrombocytopenia & renal failure were, in part, study-drug related. AI411- Pulmonary aspergillosis. Died on Relapsed AML. 5 131-001- therapy day 37 (CFTAZ 2gIVq8h). 034 AI411- Respiratory failure/AML. Died 11 Differential diagnosis for respiratory failure 11 131-001- days post therapy (CFTAZ 2gIVq8h). includes pneumonia, heart failure, 066 leukemic infiltrates. Refused mechanical ventilation. Cannot rule out study drug failure, though additional factors may have contributed to the patient's death.

AI411- Disseminated Aspergillosis. Died on AML, severe neutropenia, Prior HX of 5 131-001- therapy day 25 (CFTAZ 2gIVq8h). Aspergillus infection of left orbit & sinuses. 072 AI411- Veno-occlusive disease/multi-organ AML s/p autologous bone marrow 1 131-001- failure. Died on therapy day 28 transplant. [Also enrolled as 411-131- 089 (CFTAZ 2gIVq8h). 001-076, received Ceftazidime] AI411- Hyponatremia, seizure. Died 30 days [Also enrolled as 411-131-003-008, 1 131-003- post therapy (CFTAZ 1.1gmIVq12h). received Ceftazidime] 020

Table 8. M.O. Likely Etiologies and Contributing Factors for Patient Deaths < 30 Days Post Study Therapy in Febrile Neutropenia Studies (cont'd)

Study AI411-137

COMPARATOR TREATMENT ARM [gentamicin (GENT) & mezlocillin (MZN)] Patient Likely Cause of Death per M.O. Major Contributing Factor(s) per M.O. CODE ID AI411- CMV pneumonia/ARDS/Veno- Alveolar hemorrhage, Hodgkin's disease, 7A 137-001- occlusive liver disease. Died on GI bleed, renal failure. 031 therapy day 25 of MZN & post therapy day 21 of GENT.

Table 8. M.O. Likely Etiologies and Contributing Factors for Patient Deaths < 30 Days Post Study Therapy in Febrile Neutropenia Studies (cont'd)

Study AI411-186

CEFEPIME (CFP) TREATMENT ARM (+ Amikacin) Patient Likely Cause of Death Major Contributing Factor(s) CODE ID AI411- E. cloacae sepsis, pneumonia. Died Neutropenia, acute leukemia. 7B 186-002- 29 days post therapy (CFP 2gIVq12h 006 + amikacin q12h). AI411- Cerebral hemorrhage. Died 2 days Thrombocytopenia, chemotherapy, acute 6 186-002- post therapy (CFP 2gIVq12h + leukemia. Cannot completely rule out 007 amikacin q12h). possibility that study therapy may have contributed to thrombocytopenia. AI411- RSV pneumonia. Died 29 days post Pancytopenia, acute Non-Hodgkin's 7A 186-004- CFP 2gIVq12h, 31 days post Lymphoma. Cannot rule out that study 006 amikacin q12h. therapy contributed to renal failure. AI411- S. aureus (MSSA) septic shock. Died Acute lymphoblastic leukemia. Cannot 11 186-004- 2 days post single dose study therapy rule out study therapy failure, though 007 (CFP 2gIVq12h + amikacin q12h). only received single dose of cefepime.

AI411- Respiratory failure. Died 8 days post Disseminated Aspergillosis of lung & 5 186-004- CFP 2gIVq12h, 11 days post brain, Non-Hodgkin's Disease. 021 amikacin q12h. AI411- S. mitis sepsis (resistant to study ARDS/multiple organ failure. Acute 4A 186-015- therapy). Died 17 days post CFP leukemia. 004 2gIVq12h, 2 days post amikacin q12h. AI411- Acute hepatic failure with Acute leukemia with severe aplasia, renal 11 186-020- encephalopathy. Died 4 days post failure, alveolar hemorrhage, possible 004 CFP 2gIVq12h, 2 days post amikacin intra-abdominal infection. Cannot rule q12h. out study therapy failure, though additional factors may have contributed to the patient's death, or that study therapy contributed to hepatic & renal failure.

AI411- Aspergillosis (brain & pulmonary). Acute leukemia. Cannot rule out study 5 186-020- Died 6 days post CFP 2gIVq12h, 13 therapy contributing to renal failure. 011 days post amikacin q12h. AI411- Multi-organ failure (MOF). Died 2 Etiology for MOF may include: failed bone 11 186-026- days post CFP 2gIVq12h, 5 days post marrow transplant (BMT), acute leukemia, 009 amikacin q12h. sepsis due to unknown pathogen, and drug toxicity. Cannot rule out study therapy failure, though additional factors may have contributed to the patient's death.

AI411- Multi-organ failure (MOF). Died 14 Etiology for MOF may include: sepsis due 11 186-026- days post CFP 2gIVq12h, 20 days to unknown pathogen, failed bone marrow 011 post amikacin q12h. transplant (BMT), acute leukemia, drug toxicity. Cannot rule out study therapy failure, though additional factors may have contributed to the patient's death. Cannot rule out serotonin syndrome.

AI411- Refractory acute myelogenous 1 186-028- leukemia. Died 1 day post therapy 001 (CFP 2gIVq12h + amikacin q12h). AI411- Refractory acute myelogenous 1 186-028- leukemia. Died 20 days post therapy 012 (CFP 2gIVq12h + amikacin q12h). AI411- Veno-occlusive disease. Died 24 Cannot rule out that initial study therapy 1 186-029- days post CFP 2gIVq12h, 20 days failed; however patient did not appear to 002 post amikacin q12h. ultimately die from the septic event. Cannot rule out that study therapy contributed to hepatic & renal failure

Table 8. M.O. Likely Etiologies and Contributing Factors for Patient Deaths < 30 Days Post Study Therapy in Febrile Neutropenia Studies (cont'd)

Study AI411-186

COMPARATOR TREATMENT ARM [ceftazidime (CFTAZ) & amikacin (AMK)] AI411- Lymphoma. Died 22 days post 1 186-004- CFTAZ & 8 days post AMK. 019 AI411- Lymphoma. Died 14 days post Comatose 4 days prior to death. 1 186-011- CFTAZ & AMK. 003 AI411- MRSA pneumonia & septic shock. Non-Hodgkin's lymphoma. Study therapy 4A 186-023- Died 9 days post CFTAZ & on may have also contributed to the renal 001 therapy day 11 of AMK. and hepatic failure. AI411- C. albicans fungemia. Died 10 days Leukemia, possible MRSA sepsis, cannot 5 186-027- post CFTAZ & 29 days post AMK. rule out that study therapy may have 007 contributed to hepatic failure.

Table 8. M.O. Likely Etiologies and Contributing Factors for Patient Deaths < 30 Days Post Study Therapy in Febrile Neutropenia Studies (cont'd)

Study AI411-189

CEFEPIME (CFP) TREATMENT ARM Patient Likely Cause of Death Major Contributing Factor(s) CODE ID AI411- Progressive anaplastic carcinoma. 1 189-003- Died 16 days post therapy (CFP 002 2gIVq8h). AI411- E. coli septic shock. Died on therapy Hodgkin's disease. Cannot rule out 11 189-005- day 2 (CFP 2gIVq24h). study drug failure, though was in 025 shock at baseline. AI411- Cerebral bleed secondary to AML. 6 189-009- Died 3 days post therapy (CFP 012 2gIVq8h). AI411- ARDS and Pulmonary hemorrhage. S. aureus sepsis. Cannot rule out study 11 189-010- Died on therapy day 1 (CFP 2gIV x 1 drug failure, though only received 1 032 dose). dose of study therapy. AI411- Pneumonia. Died 4 days post therapy Hodgkin's disease, acute renal failure. 2 189-018- (CFP 2gIVq8h). Cannot rule out study drug failure. 010 Cannot rule out study drug nephrotoxicity. AI411- Acute lymphoblastic leukemia. Died 1 189-019- 12 days post therapy (CFP 2gIVq8h). 011 AI411- Persistent pneumonia, progressive Cannot rule out study drug failure. 2 189-022- multiple myeloma. Died 4 days post 001 therapy (CFP 2gIVq8h). AI411- End stage CLL. Died 14 days post Heart failure. 1 189-022- therapy (CFP 2gIVq8h). 013

Table 8. M.O. Likely Etiologies and Contributing Factors for Patient Deaths < 30 Days Post Study Therapy in Febrile Neutropenia Studies (cont'd)

Study AI411-189

COMPARATOR TREATMENT ARM [ceftazidime (CFTAZ)] Patient Likely Cause of Death per M.O. Major Contributing Factor(s) per M.O. CODE ID AI411- Acute GI bleed. Died 12 days post Thrombocytopenia secondary to AML & 1 189-006- therapy (CFTAZ 2gIVq8h). chemotherapy. Cannot rule out initial 014 study drug failure; however not the ultimate cause of death. AI411- Complications of liver biopsy, [Also enrolled as 411-189-007-010, 1 189-007- relapsed lymphoma. Died 5 days post received Ceftazidime] 015 therapy (CFTAZ 2gIVq8h). AI411- Hemorrhagic shock, pericardial Thrombocytopenia, heparin, PCP 1 189-009- tamponade. Died 21 days post pneumonia. 041 therapy (CFTAZ 2gIVq8h). AI411- Cardiogenic shock/dilated Renal & hepatic failure, Enterococcal 1 189-009- cardiomyopathy. Died 8 days post bacteremia & UTI. 044 therapy (CFTAZ dose not provided). AI411- Progressive lymphoma. Died 22 days 1 189-010- post therapy (CFTAZ 2gIVq8h). 026 AI411- ARDS. Died 8 days post therapy Possible pneumonia. AML. Progressive 11 189-010- (CFTAZ 2gIVq8h). multi-organ failure. Cannot rule out 027 study drug failure, though additional factors may have contributed to the patient's death.

AI411- ARDS/pneumonia due to M. AML, neutropenia post chemotherapy. 4A 189-016- sedentarius. Died 11 days post 005 therapy (CFTAZ 2gIVq8h). AI411- HSV pneumonia. Died 25 days post NHL. 7A 189-016- therapy (CFTAZ 2gIVq8h). 011 AI411- Uremia secondary to multiple 1 189-022- myeloma. Died 16 days post therapy 003 (CFTAZ 2gIVq8h). AI411- Unclear etiology. Died on therapy Differential diagnosis: overwhelming 11 189-023- Day 2 (CFTAZ 2gIVq8h). sepsis, AML in blast crisis, liver failure, 001 less likely anaphylactic reaction to study therapy. Cannot rule out study drug failure, though additional factors may have contributed to the patient's death.

Table 8. M.O. Likely Etiologies and Contributing Factors for Patient Deaths < 30 Days Post Study Therapy in Febrile Neutropenia Studies (cont'd)

Study AI411-198 CEFEPIME (CFP) TREATMENT ARM [+ Vancomycin (VANCO)] Patient Likely Cause of Death Major Contributing Factor(s) CODE ID AI411- S. sanguis sepsis. Died on therapy Cannot rule out study drug failure. 2 198-002- day 4 (CFP 2gIVq8h+VANCO). ARDS, Acute leukemia. [Also enrolled as 018 411-198-002-012, received Cefepime] AI411- Refractory leukemia. Died 14 days Placed on home hospice, pneumonitis (Cx 1 198-002- post therapy (CFP negative). 026 2gIVq8h+VANCO). AI411- Stomatococcus mucilaginosus Hodgkin's lymphoma. Pathogen was 4A 198-002- sepsis, multiple organ failure. Died 1 resistant to cefepime, therefore not 029 day post therapy (CFP considered failure due to cefepime. 2gIVq8h+VANCO). AI411- Cerebral hemorrhage. Died on Thrombocytopenia secondary to 6 198-004- therapy day 10 (CFP chemotherapy & leukemia. Cannot rule 013 2gIVq8h+VANCO). out that cefepime may have contributed to thrombocytopenia. AI411- Streptococcus oralis sepsis. Died 18 Cannot rule out study drug failure, 11 198-004- days post therapy (CFP though additional factors may have 029 2gIVq8h+VANCO). contributed to the patient's death. Pneumonia/ARDS. Acute leukemia.

Table 8. M.O. Likely Etiologies and Contributing Factors for Patient Deaths < 30 Days Post Study Therapy in Febrile Neutropenia Studies (cont'd)

Study AI411-198

COMPARATOR TREATMENT ARM [ceftazidime (CFTAZ) & vancomycin (VANCO)] Patient Likely Cause of Death Major Contributing Factor(s) CODE ID AI411- Pneumonia. Died 1 day post CFTAZ Cannot rule out study drug failure, 11 198-001- 2gIVq8h & 9 days post VANCO though additional factors may have 003 IVq8h. contributed to the patient's death. Renal insufficiency, acute leukemia. AI411- Septic shock/multiple organ failure. Pathogen unknown, renal failure. 11 198-001- Died 1 day post CFTAZ 2gIVq8h & 2 Lymphoma. Cannot rule out study drug 014 days post VANCO IVq8h. failure, though additional factors may have contributed to the patient's death. AI411- Pneumonitis/ARDS/shock. Died on Lymphoma. Pathogen unknown. Cannot 11 198-002- therapy day 10 of CFTAZ 2gIVq8h & rule out study drug failure, though 005 7 days post VANCO. additional factors may have contributed to the patient's death.

Table 8. M.O. Likely Etiologies and Contributing Factors for Patient Deaths < 30 Days Post Study Therapy in Febrile Neutropenia Studies (cont'd)

Study AI411-204 CEFEPIME (CFP) TREATMENT ARM Patient Likely Cause of Death Major Contributing Factor(s) CODE ID AI411- E. coli septic shock. Died on therapy Metastatic lung cancer. Cannot rule out 11 204-003- day 2 (CFP 2gIVq24h). study drug failure, though additional 042 factors may have contributed to the patient's death. AI411- Aspiration pneumonia. Died 8 days Pulmonary edema, multiple myeloma. 1 204-003- post therapy (CFP 2gIVq12h). 341 AI411- Sequelae of metastatic breast cancer. 1 204-005- Died 15 days post therapy (CFP 142 2gIVq8h). AI411- Sequelae of chronic myelogenous 1 204-005- leukemia in blast crisis, persistent 335 neutropenia. Died 21 days post therapy (CFP 2gIVq8h). AI411- Coagulase-negative Staphylococcal Acute leukemia, multiple myeloma with 4A 204-005- bacteremia/sepsis. Died 6 days post CNS involvement, persistent neutropenia. 472 therapy (CFP 2gIVq8h). AI411- E. faecium sepsis, cardiac Hyperkalemia, acute renal failure, chronic 4A 204-005- arrhythmia. Died 9 days post therapy myelogenous leukemia in blast crisis. 475 (CFP 2gIVq8h). AI411- T. glabrata fungemia. Died 17 days Acute lymphocytic leukemia, progressive 5 204-005- post therapy (CFP 2gIVq8h). renal dysfunction, metabolic acidosis, 559 possible pneumonia. AI411- Adenocarcinoma with brain Experienced grand mal seizures approx. 7 1 204-007- metastases. Died 26 days post days post therapy attributed to brain 022 therapy. (CFP 2gIVq8h). metastases. AI411- Advanced lymphoma. Died 21 days 1 204-010- post therapy (CFP 2gIVq8h). 070 AI411- Respiratory failure due to advanced 1 204-010- lung cancer, family withdrew care. 133 Died 7 days post therapy (CFP 2gIVq8h). AI411- Advanced metastatic breast 1 204-010- carcinoma on hospice care. Died 20 186 days post therapy (CFP 2gIVq8h). AI411- Disseminated lymphoma, family Septic shock (K. pneumoniae bacteremia 11 204-010- withdrew care. Died 9 days post & E. aerogenes UTI); renal failure. 366 therapy (only received 1 dose of 2g Cannot rule out study drug failure, CFP, then switched to alternative though additional factors may have therapy within hours of enrollment contributed to the patient's death. due to septic shock).

AI411- C. krusei sepsis. Died 22 days post Acute myelogenous leukemia. 5 204-010- therapy (CFP 2gIVq8h). 561 AI411- Acute leukemia. Died 22 days post 1 204-012- therapy (CFP 2gIVq8h).

373

AI411- Esophageal cancer, GI bleed, family Aspiration pneumonia (C. freundii), 1 204-014- withdrew care. Died 7 days post radiation-induced esophageal necrosis. 117 therapy (CFP 2gIVq8h). AI411- Withdrawn from life support due to Small cell lung cancer. 1 204-016- semi-comatose state secondary to fall 102 sustained in hospital. Died 9 days post therapy (CFP 2gIVq12h). AI411- Aspergillus pneumonia. Withdrawn Non-Hodgkin's lymphoma, pancytopenia. 5 204-016- from life support. Died 11 days post 402 therapy (CFP 2gIVq8h). AI411- Cerebral hemorrhage. Died 14 days Thrombocytopenia, chronic myelogenous 6 204-016- post therapy (CFP 2gIVq8h). leukemia in blast crisis. 455 AI411- Left basal ganglion mass. Died 5 Recurrent acute leukemia. 1 204-016- days post therapy (CFP 2gIVq8h). 490 AI411- Respiratory failure from Acute myelogenous leukemia with 1 204-016- chemotherapy-induced lung injury. persistent neutropenia. 541 Died 25 days post therapy (CFP 2gIVq8h). AI411- Respiratory failure. Died 23 days post Metastatic breast cancer. 1 204-018- therapy (CFP 2gIVq8h). 233

Table 8. M.O. Likely Etiologies and Contributing Factors for Patient Deaths < 30 Days Post Study Therapy in Febrile Neutropenia Studies (cont'd)

Study AI411-204 COMPARATOR TREATMENT ARM [ceftazidime (CFTAZ)] Patient Likely Cause of Death per M.O. Major Contributing Factor(s) per M.O. CODE ID AI411- Lymphoma. Died 25 days post 1 204-005- therapy (CFTAZ 2gIVq8h). 395 AI411- E. cloacae & K. pneumoniae septic ALL, secondary Staphylococcal sp. 2 204-005- shock. Died 4 days post therapy bacteremia. Cannot rule out study drug 471 (CFTAZ 2gIVq8h). failure. [Also enrolled as AI411-204- 005-393, received CEFEPIME] AI411- C. guillermondii fungal sepsis. Died Prolonged neutropenia, AML. 5 204-005- 13 days post therapy (CFTAZ 491 2gIVq8h). AI411- Metastatic GI adenocarcinoma. Died 1 204-010- 6 days post therapy (CFTAZ 067 2gIVq8h). AI411- Clostridium septicum septic shock. Esophageal adenocarcinoma with liver 2 204-010- Died 18 days post therapy (CFTAZ metastases. Cannot rule out study drug 135 2gIVq8h). failure. [Also enrolled as AI411-204- 010-132, received CEFEPIME] AI411- Metastatic breast cancer. Died 13 1 204-010- days post therapy (CFTAZ 2gIVq8h). 188 AI411- Lymphoma. Died 10 days post 1 204-010- therapy (CFTAZ 2gIVq8h). 189 AI411- C. krusei line sepsis. Died 9 days CML in blast crisis. 5 204-010- post therapy (CFTAZ 2gIVq8h). 369 AI411- Cardiac arrest. Died 1 day post Metastatic breast cancer. 1 204-013- therapy (CFTAZ 2gIVq8h). 161 AI411- Respiratory failure and pulmonary Acute AML and myelodysplastic 1 204-016- hemorrhage. Died 7 days post syndrome. 406 therapy (CFTAZ 2gIV x 1 dose). AI411- Leukemia. Died 2 days post therapy Hodgkin's lymphoma. [Also enrolled as 1 204-016- (CFTAZ 2gIVq8h). AI411-204-016-401, received 484 CEFEPIME] AI411- S. aureus (MSSA) sepsis. Died 7 Cannot rule out study drug failure or 2 204-016- days post therapy (CFTAZ 2gIVq8h) toxicity resulting in renal failure. High- 485 grade lymphoma. AI411- Progressive leukemia. Died 25 days Thrombocytopenia, vaginal hemorrhage. 1 204-017- post therapy (CFTAZ 2gIVq8h). 424

Table 8. M.O. Likely Etiologies and Contributing Factors for Patient Deaths < 30 Days Post Study Therapy in Febrile Neutropenia Studies (cont'd)

CEFEPIME NON-COMPARATIVE STUDIES

Study AI411-143 Patient Likely Cause of Death Major Contributing Factor(s) CODE ID AI411- S. pneumoniae Cannot rule out study drug failure. 2 143-001- pneumonia/bacteremia. Placed on ARDS, CHF, metastatic lung cancer. 029 comfort care after 5 days on CFP. Died on therapy day 6 (CFP 2gIVq8h).

AI411- Disseminated candidiasis, pulmonary Acute leukemia, CHF, massive pleural 5 143-001- emboli. Died 1 day post therapy (CFP effusion. [Also enrolled as AI411-143- 068 2gIVq8h). 001-059] AI411- Disseminated tuberculosis. Died at Myelodysplasia with pancytopenia. 1 143-002- home 14 days post therapy (CFP 005 2gIVq8h). AI411- Metastatic seminoma. Died 26 days Interstitial pulmonary process (bleomycin 1 143-002- post therapy (CFP 2gIVq8h). toxicity) vs. pneumonia, multiple organ 009 failure. AI411- Myelodysplasia in blast crisis. Died 1 Progressive multiple organ failure. 1 143-002- day post therapy (CFP 2gIVq8h). 017 AI411- P. aeruginosa sepsis. Died 17 days Refractory leukemia, placed on comfort 1 143-002- post therapy (CFP 2gIVq8h). care. [Also enrolled as 411-143-002- 021 015] AI411- Clostridium septicum septic shock. Widely metastatic breast cancer. 11 143-002- Died within hours of 1st dose of CFP 038 (2gIVq8h). AI411- S. aureus septic shock. Died within Cannot rule out study drug failure; 11 143-002- hours of 1st dose of CFP (2gIVq8h). however, patient had progressed to 040 septic shock prior to completion of 1st cefepime dose. Autopsy showed fungal spores/filaments in lungs. Metastatic breast cancer.

Table 8. M.O. Likely Etiologies and Contributing Factors for Patient Deaths < 30 Days Post Study Therapy in Febrile Neutropenia Studies (cont'd)

CEFEPIME NON-COMPARATIVE STUDIES

Study AI411-158

Patient Likely Cause of Death Major Contributing Factor(s) CODE ID AI411- Disseminated candidiasis. Died 5 CML. Cannot rule out that study therapy 5 158-001- days post therapy (CFP 2gIVq8h). may have contributed to hepatic failure. 001 AI411- Hepatic failure. Died 2 days post Hepatic failure may have been due to: 1 158-001- therapy (CFP 2gIVq8h). veno-occlusive dz s/p BMT/busulfan 011 toxicity. Cannot completely rule out that cefepime may have contributed to hepatic failure.

AI411- Multi-organ failure/comfort measures. Renal failure (Diff Dx: cyclosporine vs 1 158-001- Died 2 days post 2nd course of cefepime). Aplasia (Diff Dx: failed BMT vs 017 therapy (CFP 2gIVq8h). leukemia vs cefepime). Pulmonary hemorrhage (Diff Dx: barotrauma & thrombocytopenia vs leukemia vs pneumonia (bacterial vs viral vs fungal vs opportunistic pathogen).

AI411- Hemorrhagic cystitis, bone marrow Failed BMT s/p CML. Autopsy showed 1 158-001- aplasia. Died 1 day post 2nd course bone marrow hypoplasia, hemorrhagic 025 of therapy (CFP 2gIVq8h). cystitis, and pulmonary edema.

Discussion of Death Narratives:

Study AI411-118 Cefepime Patients:

Patient AI411-118-001-020 was a 37 yo White male from the USA with refractory, metastatic Hodgkin’s lymphoma. Recent radiation therapy occurred from 12/89 to 3/90. Recent chemotherapy (6/89) included: Adriamycin, VP-16, and Cytoxan. On 4/2/90, he developed febrile neutropenia (Temp=103 F, ANC=964) and received cefepime 2gIVq8h for 9 days (4/2 4/10). Baseline blood and urine cultures and CXR were negative. Pertinent Med Hx included: melanoma resected in 1985 and chronic pancytopenia. Pertinent concomitant therapy included: danazol, Halcion, Tylenol, and prednisone. The patient became afebrile on Day 3. Cefepime was discontinued on (b) (6) and the patient was discharged. As an outpatient, he complained of progressive dyspnea. On (b) (6) , he presented to the ER with fever and hypoxia requiring intubation. He was empirically placed on Bactrim. The patient suffered a cardiac arrest and died on(b) (6) days post therapy). No autopsy was performed. The investigator deemed that the patient’s death was due to hypoxia secondary to radiation pneumonitis. The following SAE was noted in the CRF: “death” ( (b) (6) ) was deemed unrelated to cefepime.

M.O. comment: This patient’s death was due to a cardiac arrest secondary to hypoxia and most likely a result of radiation pneumonitis associated with therapy for the patient’s refractory Hodgkin’s disease.

Patient AI411-118-001-022 was a 68 yo White male from the USA with newly diagnosed leukemia. Recent chemotherapy (4/26-5/2/90) included: high dose Ara-C and mitoxantrone. On 4/24/90, he developed febrile neutropenia (Temp=101.7 F, ANC=56) and received cefepime 2gIVq8h for 8 days (4/25-5/2/90). Baseline blood and urine cultures were negative. CXR was reported as consistent with evidence of CHF. Pertinent Med Hx included: nephrolithiasis, myelosuppression, and CVA. Pertinent concomitant therapy included: allopurinol, temazepam, benadryl, meperidine, Reglan, Percocet, lasix, Tylenol, heparin, decadron, verapamil, mannitol, and electrolyte supplementation. Pertinent laboratory values included: platelets=85,000 on 4/24 and 25,000 on 5/3. On Day 2, the patient began chemotherapy with Ara-C and received lasix for CHF. The patient remained febrile and on 5/2 cefepime was discontinued and he was placed on imipenem. On 5/3, his platelet count decreased to 25,000 and he suffered an intracerebral bleed. On(b) (6) days post therapy), he sustained a myocardial infarction and suffered a cardiac arrest, electromechanical dissociation, and died. No autopsy was performed. The investigator deemed that the patient’s death was due to an acute myocardial infarction and intracranial hemorrhage. The following SAEs were noted in the CRF: “intracranial hemorrhage” (5/3 to death) and (b) “death” (6) ) were deemed unrelated to study therapy.

M.O. comment: This patient’s death was most likely due to myocardial infarction and intracranial hemorrhage. Intracranial hemorrhage was likely related to thrombocytopenia due to leukemia and subsequent chemotherapy.

Patient AI411-118-001-047 was a 22 yo Black male from the USA with acute myelogenous leukemia (AML). Recent chemotherapy (4/13-4/17, 4/27-5/1/91) included: Menogaril. On 4/26/91, he developed febrile neutropenia (Temp=104.5 F, ANC=364) and received cefepime

2gIVq8h for 28 days (4/26-5/23). Four baseline blood cultures grew E. aerogenes. Culture of the Hickman catheter site grew S. aureus. Urine culture was negative. Pertinent Med Hx included: myelosuppression, obesity, and thrush. Pertinent concomitant therapy included: Mycelex troches, PRBCs, Tylenol, meperidine, benadryl, prochlorperazine, allopurinol, Percocet, lorazepam, temazepam, and codeine. During the course of the study, the patient defervesced, blood cultures became negative, and the cellulitis at the Hickman catheter site resolved. Cefepime was discontinued on 5/23 after 28 days of therapy. On 6/1, the patient developed a new fever to 102 F and was empirically started on cefoperazone and gentamicin. Vancomycin was started for Staphylococcus species initially isolated from blood culture. He remained febrile and neutropenic. Blood cultures on 6/5, 6/6, and 6/7 grew C. albicans and amphotericin B was added. The patient developed multiple organ failure and died on (b) (6) (b) following a cardiac arrest (6) days post therapy). No autopsy was performed. The investigator deemed that the patient’s death was due to multiple organ failure, candidemia, and leukemia. The following SAEs were noted in the CRF: “death” and multisystem organ failure” (b) (6) ) were deemed unrelated to cefepime.

M.O. comment: This patient’s death was most likely due to fungal sepsis (C. albicans), multiple organ failure, and refractory leukemia.

Patient AI411-118-001-065 was a 24 yo White female from the USA with a widely metastatic neuroectodermal pituitary brain tumor. Recent chemotherapy (10/6-10/8/91) included: VP-16 and cisplatin. On 10/15/91, she developed febrile neutropenia (Temp=100.5 F, ANC not performed; however WBC=0.3) and was enrolled, however there was a 4 hour delay before she received her first and only dose of cefepime 2g IV (10/15/91 only). Physical exam revealed hypotension (SBP=82), HR=100, decreased breath sounds on right, abdominal distension, lower extremity weakness, and left sided weakness. Baseline blood culture was positive for P. aeruginosa sensitive to cefepime, piperacillin, and gentamicin. Urine culture was negative. Pertinent Med Hx included: metastatic disease to liver, ovaries (resulting in bilateral oophorectomy), and lungs (with right middle lobe compression and malignant effusion), malignant ascites, seizures related to brain tumor, and myelosuppression. Pertinent concomitant therapy included: single doses of Kefzol and vancomycin (10/8 for portacath placement), other antibacterials as noted below, platelets, decadron, G-CSF, Levophed, Dilaudid, and Tagamet. The patient received her first dose of cefepime; however her condition continued to deteriorate and within 3 hours she developed septic shock and suffered cardiopulmonary arrest. The patient was resuscitated, intubated, and antimicrobial therapy was changed to ceftazidime, vancomycin, and tobramycin. Six hours later the patient suffered another cardiopulmonary arrest and died on (b) (6) post single dose of cefepime). Autopsy revealed widespread metastatic cancer and “no immediate cause of death”. The investigator deemed that the patient’s death was caused by septic shock due to P. aeruginosa and metastatic cancer. The following SAE was noted during the course of the study: hemodynamic shock ( (b) (6) ) was not considered related to cefepime.

M.O. comment: This patient’s death was most likely due to P. aeruginosa septic shock. The M.O. cannot rule out the possibility that this represented a treatment failure attributable to cefepime; however, the M.O. also notes that, according to the CRF, there was a 4 hour delay from the time of enrollment to the time of the first dose of cefepime. Additionally, the patient

78 received a single dose of cefepime prior to being changed to a multi-antibacterial regimen that included ceftazidime.

Patient AI411-118-002-017 was a 72 yo White male from the USA with recently diagnosed adenocarcinoma of the stomach. Recent chemotherapy (12/11-1/12/90) included: 5-azacytidine. On 2/8, he developed febrile neutropenia (Temp=38.1 C, ANC=528) and received cefepime 2gIVq8h for 6 days (2/9-2/14/90). Baseline blood and urine cultures and CXR were negative. Pertinent Med Hx included: gastric ulcers and GI bleed, myelodysplastic syndrome since 1989, bilateral inguinal hernia repairs in 1962. Pertinent concomitant therapy included: PRBCs, Demerol, Pepcid, Atarax, morphine, Hismanol, and electrolyte supplementation. During the hospital course, the patient initially became afebrile from 2/9 to 2/13, and then developed a fever on 2/14 to 38.4 C. Cefepime was discontinued and the patient was placed on ampicillin, clindamycin, and tobramycin. Repeat cultures were negative. On (b) (6) , the patient underwent hemigastrectomy which was complicated post operatively by small bowel obstruction. On 2/17, repeat cultures were negative. On 2/18, sputum grew E. coli sensitive to cefepime and (b) gentamicin. On 2/21, surgical wound grew coagulase negative Staphylococci. On (6) , the patient suffered a cardiopulmonary arrest and died. No autopsy was performed. The investigator deemed that the patient’s death was due to cardiopulmonary arrest, adenocarcinoma of the stomach, small bowel obstruction, and possible intra-abdominal infection. The following SAE was noted in the CRF: “death” was deemed as unexpected, but unrelated to cefepime.

M.O. comment: This patient’s death was due to cardiopulmonary arrest status post complicated post-operative course following gastrectomy for gastric adenocarcinoma. Study drug failure was less likely given that patient was afebrile for 5 days on cefepime, repeat (b) cultures were negative, and the patient died (6) days post study therapy. The patient may have had an intra-abdominal infection or pneumonia post cefepime therapy.

The CRF for the following patient was not provided by BMS because the patient died >30 days after the last dose of cefepime. The following was based on information found in the Death Narrative. Patient AI411-118-001-007 was a 76 yo White male from the USA with lymphoma. Recent chemotherapy was not provided in the narrative. On 11/24, he developed febrile neutropenia (Temp=101.8 F, ANC unknown, total WBC=600) and received cefepime 2gIVq8h for 14 days (11/24-12/7/89). Baseline blood cultures grew K. pneumoniae. Urine culture and CXR were negative. Pertinent Med Hx included: lymphoma, atrial fibrillation, hypothyroidism, and nephrectomy in 1989. Pertinent concomitant therapy was not provided in the narrative. The patient became afebrile by Day 2. However, on Day 4 (11/27), the patient became hypotensive (SBP=70) which reportedly resolved with IV fluid hydration. On 11/30, the patient was diagnosed with C. difficile infection and began therapy with oral vancomycin. On 12/4, he developed a mental status change. CT scan and CSF were negative for an etiology and the confusion apparently resolved without additional therapy. Cefepime was discontinued on 12/7 because the patient was afebrile and had negative blood cultures. The following day, the patient’s BUN and creatinine rose to 39 and 2.8 mg/dL, respectively. IV hydration was provided and the BUN and creatinine improved. The patient developed progressive deterioration over the next several days and became unresponsive. He expired on (b) (6) days post therapy). Autopsy status was unknown. The investigator deemed that the patient’s death was due to his underlying lymphoma. SAEs were not listed in BMS’s narrative.

M.O. comment: This patient’s death was most likely due to underlying lymphoma. The patient died > 30 days post cefepime therapy.

Study AI411-118 (Comparator = piperacillin/gentamicin):

Patient AI411-118-001-011 was a 57 yo White female from the USA with relapsed Non Hodgkin’s lymphoma. Recent chemotherapy (12/15-12/19/89) included: Cytoxan, Adriamycin, and VP-16. On 12/23/89, she developed febrile neutropenia (Temp=105.6 F, ANC=89) and received piperacillin 18 g for 6 days (12/24-12/29) and gentamicin 270 mg daily for 4 days (12/23-12/26). At the time of enrollment, the patient was hypotensive (90/50). Baseline blood cultures grew coagulase-negative staphylococci. Pertinent Med Hx included: NHL had relapsed in 12/88 and 12/89, myelosuppression, resected adenocarcinoma of lung in 1976, and a 50 pack/yr smoking history. Pertinent concomitant therapy included: cefoxitin (12/13 only), vancomycin (12/25-12/19), imipenem (12/29), tazobactam (12/24-12/26), aztreonam (12/29-1/3), Flagyl (12/29-1/3), Ativan, allopurinol, Tylenol, compazine, zantac, Levophed, dopamine, folate, and kaopectate. The patient remained hypotensive and was transferred to the ICU on 12/24 and started on vasopressors. Due to a medication administration error, the patient received piperacillin/tazobactam from 12/24-26. Repeat cultures were negative on 12/24. Vancomycin was added on 12/25. On 12/26, bilirubin was noted to have increased from 0.9 to 13 mg/dL and creatinine from 0.7 to 2.2mg/dL. At that point, gentamicin was discontinued and piperacillin was changed to piperacillin/tazobactam. Pressors were discontinued (date not specified) because blood pressure was stable though patient was still febrile. On 12/28, hypotension recurred and the patient was acidotic and anuric and was started on peritoneal hemodialysis. On 12/29, imipenem was substituted for piperacillin and IV Flagyl and aztreonam were started. The subject died of multiple organ failure on (b) (6) days post combined comparator therapy). Autopsy revealed (1) pneumonia of the right lower lobe, left lower lobe, and left upper lobe; (2) a large necrotic pancreatic mass; (3) lymphoma in the omentum and periaortic lymph nodes; (4) large gastric ulcers; (5) bile tinged ascites; (6) extrinsic compression of the gall bladder by the peri-pancreatic mass; and (7) acute cystitis. The investigator deemed that the patient died due to staphylococcal bacteremia causing shock and that the death was not related to study drugs. The following SAE was noted in the CRF: “death” was deemed unrelated to study therapy.

M.O. comment: The patient’s death was most likely due to multiple organ failure with septic shock, and ultimately due to underlying lymphoma. Given the necrotic pancreatic mass and acute cystitis, it may be more likely that the patient died due to gram-negative sepsis rather than coagulase-negative staphylococci. The M.O. cannot rule out study drug failure, though the additional factors discussed above may have also contributed to the patient's death.

Patient AI411-118-001-038 was a 63 yo Hispanic male from the USA with Non-Hodgkin’s lymphoma. Recent chemotherapy included: Cytoxan, Adriamycin, vincristine, and prednisone on 10/19/90; and Cytoxan, Adriamycin, and VP-16 on 11/10/90. On 10/30/90, he developed febrile neutropenia (Temp=38.6 C, ANC=146) and received piperacillin 3g IV q4h and gentamicin 100 mg IV q8h for 12 days (10/30-11/10). Baseline blood and urine cultures were negative. Sputum and throat cultures were positive for Staphylococcus aureus. CXR was negative for infiltrates. Pertinent Med Hx included: pancytopenia. Pertinent concomitant

therapy included: bactrim (11/11-11/14), fluconazole, Tylenol with codeine, Tylenol, kayexalate, allopurinol, insulin, furosemide, lorazepam, and meperidine. Pertinent laboratory values included: phosphorus=3.5 on 10/30 and 6.4 on 11/10, creatinine=0.8 on 10/30 and 5.5 on 11/10, AST= 73 on 10/30 and 228 on 11/10, ALT=126 on 10/30 and 41 on 11/10, and total bilirubin=0.5 on 10/30 and 4.5 on 11/10. On Day 4, pharyngitis had improved however fever persisted and the patient noted increasing abdominal pain and lower extremity weakness. CT scan of the abdomen on 11/1/90 demonstrated worsening lymphoma. From 11/1 to 11/14, the patient progressively declined with severe hypophosphatemia, hemolysis (per Sponsor, possibly related to hypophosphatemia), increasing creatinine, worsening liver function. Piperacillin and gentamicin were discontinued on 11/10. The patient died on(b) (6) due to respiratory, liver, and renal failure. No autopsy was performed. The investigator deemed that the patient’s death was due to lymphoma in an accelerated phase and unrelated to study therapy. The following SAEs were noted in the CRF: both “renal failure” and “respiratory failure” (11/10-(b) (6) were deemed unrelated to study therapy.

M.O. comment: The patient’s multi-organ failure and death were most likely due to progression of underlying lymphoma. Cannot rule out that study therapy may have contributed to renal and hepatic failure.

Patient AI411-118-002-029 was a 57 yo White female from the USA with newly diagnosed acute myelogenous leukemia (AML). The patient had not received any recent chemotherapy prior to enrollment; however, she received the following post-enrollment (6/26-7/14/90): Ara-C and daunorubicin. On 6/21/90, she developed febrile neutropenia (Temp=38.5 C, ANC=159) and received piperacillin 3g IV q4h and gentamicin 60-80 mg IV q12h for 28 days (6/21-7/18). Baseline blood and urine cultures and CXR were negative. Pertinent Med Hx included: HTN and right mastectomy for breast CA. Pertinent concomitant therapy included: antibacterials as noted below, acyclovir, Septra, Xanax, versed, benadryl, Tylenol, allopurinol, lasix, cardizem, Amicar, Reglan, Ativan, electrolyte supplementation, PRBCs, platelets, and hydrocortisone. During the course of study therapy, fever persisted and empiric fluconazole was added 6/26. On 7/5 the patient was noted to have cellulitis and abscess around left thumb nail and culture revealed: CNS (coagulase-negative staphylococci), K. oxytoca, S. faecalis, P. aeruginosa, and C. parapsilosis. Vancomycin, amphotericin B, and metronidazole were added and fluconazole was discontinued. On 7/18, piperacillin and gentamicin were replaced with ceftazidime and tobramycin however the fever persisted. Post study therapy, the subject reportedly developed heart failure. On(b) (6) the patient was found unresponsive and unable to be resuscitated. Autopsy revealed CHF, hemorrhagic gastritis with fresh blood in the stomach, and nodular lesion in both upper lobes of the lungs "consistent with granulomas or perhaps fungal infection". The investigator deemed that the patient’s death was to GI bleed, CHF, pancytopenia, and AML. No SAEs were noted in the CRF.

M.O. comment: The patient’s death was most likely due to acute hemorrhagic gastritis and congestive heart failure, perhaps due to sequelae related to her underlying leukemia and/or chemotherapy.

Patient AI411-118-002-047 was a 46 yo White female from the USA with acute myelogenous leukemia (AML). Recent chemotherapy (5/31-6/6/91) included: daunorubicin and Ara-C. On

6/5/91, she developed febrile neutropenia (Temp=38.0 C, ANC=100) and received piperacillin 3g IV q4h and gentamicin 85 mg IV q8h for 5 days (6/5-6/9) Baseline blood and urine cultures and CXR were negative. Pertinent Med Hx included: hemorrhoids and amenorrhea. Pertinent concomitant therapy included: antimicrobials as noted below, nystatin, acyclovir, platelets, PRBCs, Ativan, Thorazine, Lomotil, Tylenol, lasix, electrolyte supplementation, allopurinol, Provera, Zofran, compazine, and Zantac. On 6/8 the patient developed diarrhea reportedly due to Blastocystis hominis and Flagyl was added. On 6/9, study therapy was replaced with imipenem (reason not provided, but presumably due to persistent fever). Flagyl was started on 6/10. On 6/14, blood cultures grew Candida sp., Trichosporon sp., and Xanthomonas maltophilia. Amphotericin B was added on 6/16 and 5-FU on 6/17, however, the patient died on (b) (6) No autopsy was performed. The investigator deemed that the patient’s death was due to AML, disseminated fungal infection, and X. maltophilia bacteremia and not related to study therapy. The following SAE was noted in the CRF: “death” (b) (6) was deemed unrelated to study therapy.

M.O. comment: The patient’s death was most likely due to disseminated fungal infection and X. maltophilia bacteremia, and underlying leukemia.

Study AI411-131 Cefepime Patients:

Patient AI411-131-001-009 was a 68 yo Black female from the USA with metastatic breast cancer and poorly differentiated metastatic uterine sarcoma. Recent chemotherapy (11/9/89) included: cyclophosphamide, Adriamycin, and 5-FU. On 11/20, she developed febrile neutropenia (Temp=38.8 C, ANC=276) and received cefepime 2gIVq8h for 10 days (11/20 11/29). Baseline blood culture grew B. fragilis, and urine culture was negative. Pertinent Med Hx included: diabetes, cataracts, total abdominal hysterectomy with bilateral salpingo oophorectomy in 1989. Pertinent concomitant therapy included: ampicillin (11/13-11/20), MS contin, glyburide, enalapril, coumadin, Tylenol, compazine, fresh frozen plasma (FFP). The patient defervesced in 48 hours and subsequent blood and urine cultures were negative. On 11/25, she was diagnosed with a posterior pharyngeal abscess that was aspirated. Culture grew P. aeruginosa which the treating team considered a contaminant. Clindamycin was empirically added on 11/26. On 11/29, the cefepime and clindamycin were discontinued because the patient had been afebrile for 72 hours and ANC had returned to normal. For an unknown reason (? long term prophylactic follow-up therapy?), the patient was placed on oral Augmentin (11/29-11/30) followed by IV Unasyn (11/30-12/9). Interval cultures were negative; however, due to her extensive metastatic disease, the patient requested to be placed on palliative care measures and “Do Not Resuscitate” (DNR) status. The patient died on (b) (6) days post therapy). No autopsy was performed. The investigator deemed that the patient’s death was caused by metastatic breast and uterine cancer. The following SAE was noted in the CRF: death (b) (6) was deemed unrelated to cefepime.

M.O. comment: This patient’s death was most likely due to metastatic breast and uterine cancer.

Patient AI411-131-001-013 was a 55 yo White female from the USA with acute myelogenous leukemia (AML). Recent chemotherapy (3/2-3/8/90) included: Ara-C and daunorubicin. On

3/7/90, she developed febrile neutropenia (Temp=38.9 C, ANC=176) and received cefepime 2gIVq8h for 14 days (3/7-3/20/90). Baseline blood and urine cultures were negative. Pertinent Med Hx included: breast cancer in 1984 treated with lumpectomy and radiation therapy, and radiation pneumonitis (1984). Pertinent concomitant therapy included: Mycelex troches, allopurinol, electrolyte supplementation, Tylenol, benadryl, Halcion, phenergan, Demerol, PRBCs, and platelets. Pertinent laboratory values included: Alkaline phosphatase=108 on 3/7, 270 on 3/16, and 506 on 3/20; AST=29 on 3/7, 54 on 3/16, and 54 on 3/20; ALT=49 on 3/7, 43 on 3/16, and 32 on 3/20; creatinine=0.7 on 3/7, 1.5 on 3/16, and 1.5 on 3/20; platelets=62 on 3/7, 19 on 3/16, and 30 on 3/20. During the hospital course, IV Vancomycin (3/10-3/20, 3/26-3/27) and amphotericin B (3/14-3/29) were added due to persistent fevers. Repeat blood cultures on 3/10 and 3/17 were negative. On 3/20, the patient was noted to have new bilateral lower lobe infiltrates on CXR, and cefepime and vancomycin were discontinued and she was started on IV imipenem (3/20-3/26), erythromycin (3/20-3/27), and pentamidine (3/20 only). On 3/21 she developed respiratory failure and required intubation. On 3/28 her mental status worsened and she was noted on CT scan to have a subarachnoid hemorrhage and possible right cerebral infarct. On 3/29, fevers continued and she developed disseminated intravascular coagulation (DIC) and further increased liver enzyme studies. On (b) (6) the patient’s family elected to remove the patient from the respirator and she died the same day. No autopsy was performed. The investigator deemed that the patient’s death was cause by sepsis (pathogen unknown), leukopenia, and AML, and unrelated to cefepime. The investigator believed that the rising alkaline phosphatase may have been related to hepatic candidiasis. The following SAE was noted in the CRF: death (b) (6) was deemed unrelated to cefepime.

M.O. comment: This patient’s death was most likely due to sepsis (unknown pathogen) with multi-organ failure, AML, and a CNS bleed. The differential diagnosis for the etiology of sepsis includes: bacterial, fungal, or other opportunistic pathogen. The M.O. cannot rule out study drug failure; however additional factors may have also been associated with the patient’s death.

Patient AI411-131-001-033 was a 43 yo White male from the USA with relapsed acute myelogenous leukemia (AML) s/p recent bone marrow transplant 6/13/90. Recent chemotherapy (6/13-7/3/90) included Ara-C and methotrexate. On 6/29/90, he developed febrile neutropenia (Temp=38.4 C, ANC<100) and received cefepime 2gIVq8h for 16 days (6/29-7/14) and then 2gq12h (due to worsening renal function) for 12 days (7/15-7/26). Baseline blood cultures grew: coagulase-negative Staphylococci (resistant to cefepime) and S. viridans (susceptible to cefepime); urine culture was negative for a pathogen. Baseline CXR was negative. Pertinent Med Hx included: fungal pneumonia 1989 and left vocal cord paralysis. Pertinent concomitant therapy included: Bactrim IV (6/29-7/12), mycostatin, nystatin, amphotericin B (7/1-7/26), PRBCs, platelets, hydrocortisone, morphine, CSA (?), Demerol, carafate, aldactone, morphine, electrolyte supplementation, Halcion, compazine, benadryl, Gammagard, Tylenol, lasix, MS contin, Atarax, nifedipine, chloral hydrate, Maalox, albumin, Bumex, Restoril, Dopamine, Flagyl, Ativan, Pentamidine (7/13 only), vancomycin (7/1-7/25), and acyclovir (6/29-7/4). Pertinent laboratory values on 6/29, 7/9, 7/19, and 7/26 included: bilirubin=0.9, 1.3, 1.1, 4.4; creatinine=1.0, 1.4, 1.2, 1.5; and BUN=32, 53, 45, 44. Intravenous vancomycin was added on 7/1 due to the presence of coagulase-negative Staphylococci on blood culture. Amphotericin B was also added on 7/1 for worsening oral candidiasis. On 7/4 the patient’s fever returned and on

7/5 he developed altered mentation which cleared after acyclovir was stopped (on 7/5). On 7/7, Flagyl was added for anaerobic coverage. On 7/9, the patient developed an erythematous, pustular rash over his entire body. On 7/10, Hickman catheter was removed due to cellulitis at the catheter site and grew C. albicans and Aspergillus fumigatus. The patient continued to deteriorate with rising BUN (45 mg/dL) and bilirubin (4.4 mg/dL), and died on (b) (6) (11 days post therapy). An autopsy was performed and revealed pulmonary aspergillosis. The investigator deemed that the patient’s death was caused by Aspergillus pneumonia and Candidal line sepsis. The following SAE was noted in the CRF: death (b) (6) was deemed unrelated to cefepime.

M.O. comment: This patient’s death was most likely due to pulmonary/disseminated Aspergillosis.

Patient AI411-131-001-055 was a 29 yo White male from the USA with relapsed acute lymphocytic leukemia (ALL). Recent chemotherapy (10/17-10/21/90) included VP-16 and methotrexate. On 10/23/90, he developed febrile neutropenia (Temp=39.1 C, ANC=400) and received cefepime 2gIVq8h for 17 days (10/23-11/8). Baseline blood and urine cultures were negative. CXR was negative. Pertinent Med Hx included: ulcerative colitis, depression, substance abuse, hypophosphatemia and DIC prior to enrollment and during the current hospitalization for his leukemia. Pertinent concomitant therapy included: acyclovir, Mycelex troches, fluconazole, ceftriaxone (11/12 to death), bactrim (11/12 to death), platelets, PRBCs, benadryl, Maalox, electrolyte supplementation, Tagamet, allopurinol, compazine, Tylenol, lasix, Demerol, Restoril, Ativan, cimetidine, Halcion, versed, phenergan, and Pamelor. On Day 4, vancomycin was added and a central venous catheter was removed because multiple repeat blood cultures grew coagulase-negative Staphylococci. Amphotericin B was added on Day 6 due to persistent fevers. On 11/8 (Study Day 17), cefepime was discontinued because the patient had been afebrile for 1 week. On 11/12 (4 days post therapy), the patient had altered mentation and CT scan revealed relapsed CNS leukemia and hydrocephalus. The patient was transferred to the ICU and ceftriaxone and Bactrim were started empirically. On 11/16, the patient was noted to be unresponsive. On (b) (6) days post therapy), the patient was extubated and died. No autopsy was performed. The investigator deemed that the patient’s death was caused by relapse of CNS leukemia. The following SAEs were noted in the CRF: “hydrocephalus causing altered mentation” (11/12 to death) and “death” (b) (6) were deemed unrelated to cefepime.

M.O. comment: This patient’s death was most likely due to relapsed CNS leukemia.

enterococci, Flagyl IV was added for presumed intra-abdominal infection, and the patient was transferred to the ICU. The patient’s fevers continued and she developed hypotension on 1/9 (b) (b) with an acute abdomen. Also on(6) (Study Day(6) the patient experienced cardiorespiratory arrest during an abdominal ultrasound and died. No autopsy was performed. The investigator deemed that the patient’s death was caused by an intra-abdominal infection and lymphoblastic lymphoma. The following SAE was noted in the CRF: “death” (b) (6) was deemed unrelated to cefepime.

M.O. comment: This patient’s death was due to cardiopulmonary arrest resulting from septic shock, most likely from an intra-abdominal focus, possibly peritonitis. The M.O. cannot rule out study drug failure, though no pathogen was isolated and the patient also had lymphoblastic lymphoma.

Patient AI411-131-001-075 was a 31 yo White male from the USA with Ewing’s sarcoma and s/p bone marrow transplant on 6/28/91. Recent chemotherapy (6/22-6/24/91) included VP-16 and Cytoxan. Recent radiotherapy occurred 6/24-6/26/91. On 6/29/91, he developed febrile neutropenia (Temp=101.5 F, ANC<200) and received cefepime 2gIVq8h for 10 days (6/29 7/8/91). Baseline blood, Hickman catheter tip, urine, and sputum cultures were negative. CXR was negative. Pertinent Med Hx included: s/p surgical debridement of sarcoma in 1/91, nephrolithiasis, and diffusely scattered folliculitis. Pertinent concomitant therapy included: bactrim DS prophylaxis, aztreonam (7/9-7/20), vancomycin (7/2-7/18), amphotericin B (7/6 7/12), acyclovir, Mycelex troches, PRBCs, platelets, Tylenol, Demerol, morphine, decadron, compazine, lasix, and electrolyte supplementation. Pertinent laboratory values included: platelets=75,000 on 6/29 and 31,000 on 7/9 (last value recorded in CRF). Fevers continued and vancomycin was added on 7/2 and amphotericin was added on 7/6. On 7/8 cefepime was discontinued due to suspected drug rash, and aztreonam was started. The patient’s respiratory status worsened over the ensuing days and on 7/17, he underwent bronchoscopy which revealed diffuse pulmonary hemorrhage. On (b) (6) days post therapy), the patient developed respiratory failure, multi-system organ failure, and died. An autopsy was performed and confirmed pulmonary hemorrhage. The investigator deemed that the patient’s death was caused by pulmonary hemorrhage, bone marrow failure/failed bone marrow transplant, multi-system organ failure, and Ewing’s sarcoma. The following SAE was noted in the CRF: “death” (b) (6) ) was deemed unrelated to cefepime.

M.O. comment: This patient’s death was due to pulmonary hemorrhage resulting in respiratory failure and multi-organ failure. The pulmonary hemorrhage was likely related to bone marrow failure/ failed bone marrow transplant. The patient had received chemotherapy for Ewing’s sarcoma.

Patient AI411-131-003-016 was a 7 yo Hispanic male from the USA with widely metastatic neuroblastoma. Recent chemotherapy (7/1-7/3/91) included VP-16 and high-dose carboplatin. On 7/16/91, she developed febrile neutropenia (Temp=104 F, ANC=0) and received cefepime 50mg/kgIVq8h for 5 days (7/16-7/20/91). Baseline blood and urine cultures were negative. Pertinent Med Hx included: pancytopenia due to prior chemotherapy. Pertinent concomitant therapy included: PRBCs, Tylenol, benadryl, and electrolyte supplementation. Pertinent laboratory values included: platelets=12,000 on 7/16 and 4,000 on 7/20. The patient

defervesced, cefepime was discontinued, and the patient was discharged on (b) (6) . On (b) (6) day post therapy), the patient experienced a massive intracranial bleed and died. No autopsy was performed. The investigator deemed that the patient’s death was caused by an intracranial bleed secondary to severe thrombocytopenia due to chemotherapy for metastatic neuroblastoma. The following SAEs were noted in the CRF: “death” (b) (6) ) was deemed unrelated to cefepime.

M.O. comment: This patient’s death was most likely due to intracranial hemorrhage secondary to chemotherapy-induced severe thrombocytopenia. The patient’s underlying disease was metastatic neuroblastoma.

Patient AI411-131-003-053 was an 8 month old White male from the USA with acute myeloid leukemia (AML). Recent chemotherapy (1/4-1/11/92) included high-dose Ara-C, daunorubicin, and 6-TG. On 1/12/92, he developed febrile neutropenia (Temp=103.3 F, ANC=0) and received cefepime 50mg/kgIVq8h for 2 days (1/13-1/14/92). Baseline blood, urine, and CSF cultures were negative. No additional pertinent Med Hx was included in the CRF. Pertinent concomitant therapy included: bactrim (1/6-1/8), ceftazidime (1/5-1/10, 1/15), vancomycin (1/15-1/25), erythromycin (1/16-1/19), amphotericin B (1/16-1/25), tobramycin (1/19-1/25), platelets, PRBCs, electrolyte supplementation, Tylenol, benadryl, morphine, and ibuprofen suspension. Pertinent laboratory values included: platelets=16,000 on 1/13 and 11,000 on 1/14. On the second day of therapy, the patient was found to be not eligible to continue in the study due to being < 2 years old. Therefore, cefepime was discontinued and ceftazidime was started. Over the next several days, the patient progressively deteriorated with respiratory distress and failure due to RSV pneumonia. On (b) (6) , the patient died. An autopsy was performed and revealed acute myeloid leukemia, respiratory syncytial virus (RSV) pneumonia, adult-type respiratory distress syndrome, massive diffuse fungal (C. albicans) colitis, a lung tissue culture revealed C. albicans and blood cultures revealed P. aeruginosa, enterococci, and coagulase-negative Staphylococci. The investigator deemed that the patient’s death was caused by adult respiratory distress syndrome/RSV pneumonia and acute leukemia. The following SAEs were noted in the CRF: “pseudomembranous colitis” (1/23-1/25), “ARDS” (1/23-1/25), and “death” (b) (6) were deemed unrelated to cefepime.

M.O. comment: This patient’s death was most likely due to ARDS/RSV pneumonia. The patient’s underlying disease was acute leukemia.

Patient AI411-131-003-057 [a.k.a. 411-131-003-019] was an 11 yo Asian male from the USA with acute lymphocytic leukemia (ALL). Recent chemotherapy (1/31-2/11/92) included doxorubicin, vincristine, asparaginase, and intrathecally: methotrexate, high-dose cortisone, and high-dose Ara-C. On 2/5/92, he developed febrile neutropenia (Temp= 39.6 C, ANC=0) and received cefepime 50mg/kgIVq8h for 5 days (2/5-2/9). Baseline blood and urine cultures were negative. Pertinent Med Hx included: Strep pharyngitis and prior rectal abscess. Pertinent concomitant therapy included: pentamidine, platelets, PRBCs, electrolyte supplementation, benadryl, nystatin oral suspension, and versed. The patient defervesced on cefepime and was discharged (b) (6) days post therapy), the patient “did not feel well” and was readmitted to the pediatric ICU where he suffered a cardiopulmonary arrest. Resuscitation was successful; however he suffered two additional arrests and died on(b) (6) days post therapy).

Blood and urine cultures were positive for Stomatococcus mucilaginosus. ANC was 0 at the time of readmission. No autopsy was performed. The investigator deemed that the patient’s death was due to Stomatococcus mucilaginosus sepsis and ALL. The following SAEs were noted in the CRF: “cardio-pulmonary arrest” (b) (6) ) and “death” (b) (6) were deemed unrelated to cefepime.

M.O. comment: This patient’s death was due to cardiopulmonary arrest related to Stomatococcus mucilaginosus sepsis. This profoundly neutropenic patient became septic 4 days post therapy. S. mucilaginosus is typically considered normal oral flora; however, in profoundly neutropenic patients, it has been known to cause infection. S. mucilaginosus has also been reported to be multi-drug resistant, including to cephalosporins. Therefore, it is less likely that the patient’s death resulted from study drug failure.

Study AI411-131 (Comparator = ceftazidime 2g IV q8h):

Patient AI411-131-001-016 was a 62 yo Black female from the USA with chronic lymphocytic leukemia (CLL). Recent chemotherapy (4/4/90) included vincristine and prednisone. On 3/27/90, she developed febrile neutropenia (Temp=38.5 F, ANC=696) and received ceftazidime 2gIVq8h for 28 days (3/27-4/23). Baseline blood and urine cultures were negative. CXR was negative. Pertinent Med Hx included: CLL since 1985, diabetes, heart disease, and a heavy smoking history. Pertinent concomitant therapy included: vancomycin (4/1 to death), amphotericin B (4/7 to death), acyclovir, folic acid, Diabeta, heparin, Isordil, iron, albuterol, lasix, nystatin, insulin, Tylenol, prednisone, Tylenol #3, clonidine, phenobarbitol, valium, vitamin K, Demerol, versed, nifedipine, dilantin, ranitidine, Solumedrol, and decadron. Pertinent laboratory values included: platelets=103,000 on 3/27 and 69,000 on 4/23; creatinine-0.9 on 3/27, 2.4 on 4/13, and 1.2 on 4/23. On 4/1 vancomycin was added because the patient remained febrile and neutropenic. There was also concern that the fever may have been due to her CLL and vincristine and prednisone were administered on 4/4. On 4/6, the patient's mental status deteriorated and she was found to have a large subdural hematoma (platelet count=71,000) requiring emergent neurosurgical evacuation. Fevers continued and amphotericin B was added on 4/7. Over the ensuing days, the patient remained comatose with intermittent seizures, renal function worsened, and further CNS bleeding was noted. On 4/16, the patient's status was changed to DNR (Do Not Resuscitate). On (b) (6) the patient was found to have absent respirations and blood pressure and was pronounced dead. No autopsy was performed. The investigator deemed that the patient's death was most likely due to subdural hematoma, coma and hypotension, and unrelated to study therapy. The following SAEs were noted in the CRF: “hypotension” (4/9-4/11), “comatose” (4/13 to death), “subdural hematoma” (4/6 to death), and “seizures” (4/11 to 4/13). All were deemed unrelated to ceftazidime, except for “seizure” which was deemed to have an unknown relation to study drug.

M.O. comment: The patient’s death was most likely due to CNS hemorrhages secondary to thrombocytopenia related to her underlying CLL and associated chemotherapy. The M.O. cannot rule out the possibility that worsening renal failure and thrombocytopenia were related to study therapy.

Patient AI411-131-001-034 was a 25 yo Black female from the USA with relapsed acute myelogenous leukemia (AML). Recent chemotherapy (6/30-7/25/90) included high-dose Ara-C, VP16, and mitoxantrone. On 7/11, she developed febrile neutropenia (Temp=38.6 C, ANC=200) and received ceftazidime 2gIVq8h for 30 days (7/11-8/9). Baseline blood cultures grew S. epidermidis. Urine culture grew Corynebacterium sp. CXR was negative. No additional pertinent Med Hx was included in the CRF. Pertinent concomitant therapy included: gentamicin (7/1-7/4), ampicillin (7/1-7/4), Flagyl (7/20 to death), vancomycin (7/13 to death), amphotericin B (7/23 to death), acyclovir, nystatin, PRBCs, platelets, Mycelex troches, Orthonovum, Tylenol, benadryl, allopurinol, Demerol, and electrolyte supplementation. The patient’s fever persisted until vancomycin was added on 7/13. On 7/21 bone marrow biopsy revealed residual leukemia and chemotherapy was reinitiated. Fever recurred on 7/23 and amphotericin B was added. Repeat blood cultures were negative. In early August, the patient developed increasing dyspnea and the CXR evolved to demonstrate bilateral fluffy infiltrates and a right upper lobe cavity. The patient was diagnosed with pulmonary aspergillosis on 8/11. The patient continued to decline and died on (b) (6) No autopsy was performed. The investigator deemed that the patient's death was due to pulmonary aspergillosis from underlying immune suppression related to AML, and not related to study drug. The following SAE was noted in the CRF: “death” (b) (6) ) was deemed unrelated to study therapy.

M.O. comment: The patient’s death was most likely due to pulmonary aspergillosis in the setting of relapsed acute leukemia.

Patient AI411-131-001-066 was a 63 yo White male from the USA with acute myelogenous leukemia (AML). Recent chemotherapy (4/4-4/10) included: Ara-C and daunorubicin. On 4/18/91, he developed febrile neutropenia (Temp=39.8 C, ANC=160) and received ceftazidime 2gIVq8h for 6 days (4/18-4/23). Baseline blood and catheter tip cultures grew Corynebacterium sp. CXR was negative. Pertinent Med Hx included: aplastic anemia since 1977, dental infections, peri-rectal fistula repair, right testicle removed in the 1970’s, 30 pack-year smoking history, and shellfish allergy. Pertinent concomitant therapy included: penicillin VK (4/2-4/5), Bactrim (4/11-4/14), amphotericin B (4/25 to death), vancomycin (4/24 to death), erythromycin (4/27 to death), tobramycin (4/20 to death), aztreonam (4/23 to death), mezlocillin (4/25 to death), Mycelex troches, PRBCs, platelets, benadryl, Tylenol, naprosyn, Demerol, zantac, Halcion, GM-CSF, Lasix, and electrolyte supplementation. The patient’s fever persisted and on 4/22, ceftazidime was discontinued due to suspected drug rash and the patient was started on aztreonam and then vancomycin. Repeat blood cultures on 4/22 were negative and remained negative for the rest of the hospital course. Over the next 10 days, the patient continued to deteriorate with worsening respiratory status. Mezlocillin, amphotericin B, erythromycin, and tobramycin were added. On 4/30, the patient was transferred to the intensive care unit for worsening respiratory status. CXR progressively revealed increased bilateral infiltrates and pleural effusion. Bone marrow biopsy on 4/25 revealed relapsing leukemia. He refused ventilatory support and eventually died on(b) (6) days post therapy). Autopsy status was not provided by BMS. The investigator felt that the patient died was due to respiratory failure secondary to pneumonia and AML and was not related to study therapy. The following SAEs were noted in the CRF: “bloating” (4/21 to death), “rales” (4/22 to death), “pleural effusion” (4/22 to death), and “death” (b) (6) were deemed unrelated to study therapy.

M.O. comment: The patient’s death was due to respiratory failure (differential diagnosis includes: pneumonia due to bacterial vs. fungal vs. other opportunistic pathogen; vs. congestive heart failure; vs. leukemic infiltrates) and acute leukemia. The M.O. cannot rule out study drug failure, though additional factors may have contributed to the patient's death.

Patient AI411-131-001-072 was a 72 yo White male from the USA with relapsed acute myelogenous leukemia (AML). Recent chemotherapy (6/11-6/17/91) included: Ara-C. On 6/18/91, he developed febrile neutropenia (Temp=39.0 C, ANC<200) and received ceftazidime 2gIVq8h for 25 days (6/18-7/12). Baseline blood cultures were positive for coagulase-negative Staphylococci; urine culture grew mixed organisms, and sputum culture grew E. cloacae; however CXR was negative. Pertinent Med Hx included: pancytopenia, Aspergillus infection of sinuses and left orbit, aortic aneurysm, gout, nephrolithiasis, appendectomy, otitis media, and tobacco use. Pertinent concomitant therapy included: cefotetan (6/11 only), vancomycin (6/20 to death), amphotericin B (6/22 to death), acyclovir, platelets, PRBCs, electrolyte supplementation, Tylenol, benadryl, Demerol, lasix, Lomotil, Zaroxylin, Bumex, Zantac, Narcan, Ativan, and morphine. Vancomycin was added on 6/20 due to persistent fevers. On 6/21, blood cultures grew Acinetobacter anitratus sensitive to ceftazidime. On 6/22, he began to complain of left eye pain and amphotericin B was added given his prior history of Aspergillus infection of the left orbit. On 6/26, his Hickman catheter was removed and cultured and grew S. epidermidis and enterococci sensitive to vancomycin. On 7/1, the patient developed mental status changes. Head CT revealed periorbital and sinusoidal inflammation, but no abscesses or bleed. CXR on 7/1 revealed new bilateral lower lobe infiltrates. Because of a positive HSV-Ig titer, acyclovir was started for possible HSV pneumonia. Lumbar puncture on 7/7 was unrevealing. Repeat blood cultures on 7/8 and 7/9 were negative. On 7/10, the patient (b) (6) (b) became comatose and on he was found unresponsive and pronounced dead (Day(6) on therapy). No autopsy was performed. The investigator deemed that the patient’s death was due to HSV pneumonia, sepsis, prolonged neutropenia, and AML. The following SAE was noted in the CRF: “death” (b) (6) was deemed unrelated to study therapy. Of note, BMS’s consultant noted a high suspicion for disseminated Aspergillosis as the cause of this patient’s death.

M.O. comment: Given the patient’s prior history of Aspergillus infection, the patient’s death was most likely due to disseminated Aspergillosis. This was in the setting of prolonged neutropenia/pancytopenia and acute leukemia.

Patient AI411-131-001-089 [a.k.a. AI411-131-001-076] was a 32 yo White male from the USA with relapsed acute myelogenous leukemia (AML) and s/p autologous bone marrow transplant on 9/5/91. Recent chemotherapy (8/27-9/3/91) included: busulfan, Cytoxan, and methotrexate. He also underwent total body irradiation in 8/91. On 9/9, he developed febrile neutropenia (Temp=39.8 C, ANC<200) and received ceftazidime 2gIVq8h for 28 days (9/9-10/6). Baseline blood cultures were positive for E. coli. Urine culture and CXR were negative. Pertinent Med Hx included: hemochromatosis with elevated liver enzymes, esophageal bleed in 6/91, and sepsis due to S. viridans in 6/91. Pertinent concomitant therapy included: Bactrim (8/31-9/1), vancomycin (9/21 to death), amphotericin B (9/24 to death), Mycelex troches, acyclovir, platelets, PRBCs, erythropoietin, Desferal, Reglan, dopamine, lactulose, pentamidine, DDAVP, albumin, Bumex, lasix, Demerol, morphine, benadryl, Halcion, Ondansetron, Tylenol, Ativan, aldactone, nifedipine, Tylox, and electrolyte supplementation. Pertinent laboratory values

included: bilirubin=0.6 on 9/9, 4.0 on 9/19, and 3.7 on 10/6. The patient initially defervesced and repeat blood cultures were negative; however, on 9/16 he became febrile again. Vancomycin was started on 9/21 and amphotericin B was started on 9/24. On 9/25, the patient underwent paracentesis which revealed massive bloody ascites. On 10/3, he became encephalopathic, and by 10/5 he was unresponsive and obtunded and a “do not resuscitate” order was placed. The patient died on (b) (6) on therapy). Autopsy revealed veno-occlusive disease of the abdomen and thorax, and lung parenchymal congestion with early upper lobe consolidation. The investigator deemed that the patient’s death was due to veno-occlusive disease, multi-organ failure, AML, and possible pneumonia. The following SAEs were noted in the CRF: “organ failure”(b) (6) and “death” ((b) (6) were deemed unrelated to study therapy.

M.O. comment: The patient’s death was most likely due to veno-occlusive disease and multi- organ failure. The patient’s underlying disease was AML s/p autologous bone marrow transplant.

The CRF for the following patient was not provided by BMS because the patient died >30 days after the last dose of study drug. The following was based on information found in the Death Narrative. Patient AI411-131-001-088 was a 42 yo White male from the USA with relapsed acute lymphocytic leukemia (ALL). Recent chemotherapy (early 9/91) included: Ara- C, mitoxantrone, and VP-16. On 9/7, he developed febrile neutropenia (Temp=101.5 F, ANC=unknown) and received ceftazidime 2gIVq8h for 40 days (9/7-10/16/91). Baseline cultures were negative. Additional pertinent Med Hx was not noted. Additional pertinent concomitant therapy was not provided. During the hospital course, vancomycin was added for possible Hickman catheter site infection. The patient remained afebrile from 9/20 to 10/16. The patient received a second course of chemotherapy during that period. Unfortunately the cancer did not go into remission and the patient made the decision to be discharged home because no further therapy was available. The patient died on(b) (6) months post therapy). Autopsy status was not provided. The investigator deemed that the patient’s death was due to sepsis, severe pancytopenia, and refractory ALL.

M.O. comment: The patient died beyond the 30-day window for this review. Death was reportedly due to sepsis, pancytopenia, and refractory leukemia.

Patient AI411-131-003-020 [a.k.a. AI411-131-003-008] was a 2 yo Hispanic male from the USA with rhabdomyosarcoma. Recent chemotherapy (7/22/91) included: vincristine. He also underwent radiation therapy to the bladder in 4/91-5/91. On 7/26/91, he developed neutropenia (ANC=200), but no fever (Temp=37.0 C), and received ceftazidime 1150 mg q12h for 2 days (7/27-7/28). Study drug was stopped due to ineligibility for study; no fever. Baseline blood and urine cultures were negative. No additional pertinent Med Hx was noted in the CRF. Pertinent concomitant therapy included: Bactrim (7/29-7/31), PRBCs, Benadryl, Tylenol, and electrolyte supplementation. Pertinent laboratory values included: sodium=134 on 7/26 and 140 on 7/28. The patient died on(b) (6) days post therapy). Autopsy report was not provided in the CRF. The investigator deemed that the patient’s death was due to hyponatremia and resultant cerebral edema, intracranial hypertension, and seizure. No SAEs were noted in the CRF.

M.O. comment: The patient’s death was most likely due to underlying disease/co-morbid illness: hyponatremia and resultant cerebral edema, intracranial hypertension, and seizure.

This patient died >30 days after the last dose of study drug. Patient AI411-131-003-043 was a 12 yo Hispanic male from the USA with acute lymphocytic leukemia. Recent chemotherapy (11/21/91) included: VP-16, Cytoxan, and Mesna. He also underwent radiation therapy from 10/91-11/91. On 11/27/91, he developed neutropenia (ANC=360) and fever (Temp=38.9 C), and received ceftazidime 2gIVq8h for 8 days (11/27-12/4/91). Baseline blood and urine cultures were negative. Pertinent Med Hx included rectal inflammation and suspected VZV infection. Pertinent concomitant therapy included: Bactrim, Flagyl, acyclovir, PRBCs, platelets, Benadryl, Tylenol, and electrolyte supplementation. Pertinent laboratory values included: platelets=12,000 on 11/27 and 80,000 on 12/4. The patient survived the episode of febrile neutropenia and eventually died on (b) (6) days post therapy). Autopsy report was not provided in the CRF. The investigator deemed that the patient’s death was due to acute lymphocytic leukemia. No SAEs were noted in the CRF.

M.O. comment: The patient’s death was most likely due to acute lymphocytic leukemia.

Study AI411-137 (Comparator = gentamicin + mezlocillin):

Patient AI411-137-001-031 was a 25 yo White female from the USA with recurrent Hodgkin’s lymphoma, s/p allogeneic bone marrow transplant (ABO mismatched) on 6/13/91. Recent chemotherapy (6/6-6/18/91) included: busulfan and cyclophosphamide; she had also received radiotherapy (12/88-3/89). On 6/18/91, she developed febrile neutropenia (Temp=101.4 F, ANC=120) and received mezlocillin 3gIVq4h for 25 days (6/18-7/12) and gentamicin 60-360 mg IVqd for 4 days(6/18-6/21). Baseline blood and urine cultures were negative. CXR was negative. Pertinent Med Hx included: s/p splenectomy, and HSV zoster in 1990. Pertinent concomitant therapy included: antimicrobial agents as noted below, acyclovir, Orthonovum, clotrimazole troches, Pepcid, Tylenol, Trental, Ativan, dilaudid, Halcion, oxycodone, lasix, PRBCs, and platelets. Pertinent laboratory values included: platelets=12,000 on 6/18 and 25,000 on 6/25, AST=50 on 6/20 and 569 on 6/27 and 215 on 7/5, ALT= 35 on 6/20 and 222 on 6/27 and 104 on 7/5, and total bilirubin=2.2 on 6/20 and 21.3 on 7/5. On 6/20, she began developing signs of hepatic failure with rising bilirubin, ascites, and elevated liver enzyme studies. Due to ongoing fever, tobramycin was substituted for gentamicin on 6/21. The patient reportedly defervesced by 6/23; however her overall clinical status continued to worsen and she was diagnosed with veno-occlusive disease secondary to bone marrow transplant. On 6/25 she experienced respiratory distress and diffuse alveolar hemorrhage, renal failure, and required ICU care and subsequent intubation due to ARDS. Additional antimicrobial therapy in addition to mezlocillin and tobramycin included: ceftazidime, vancomycin, Bactrim, and Flagyl. The (b) (6) (b) (b) patient died on (Day(6) of mezlocillin and(6) days post gentamicin). Autopsy revealed nodular sclerosing Hodgkin’s disease, alveolar hemorrhage, CMV pneumonia, veno-occlusive disease of the liver, upper GI bleeding, and renal failure. The investigator deemed that the patient’s death was due to veno-occlusive disease, CMV pneumonia, pulmonary hemorrhage, and Hodgkin’s disease. The following SAEs were noted in the CRF: “Veno-occlusive Disease” (6/23-7/12), “Diffuse alveolar hemorrhage syndrome” (6/23-7/12) and “Death” (b) (6) were deemed unrelated to study therapy.

M.O. comment: This patient’s death was most likely due to a combination of co-morbid conditions linked to a failed bone marrow transplant: veno-occlusive liver disease, CMV pneumonia with pulmonary hemorrhage and ARDS; and ongoing active Hodgkin’s disease.

Study AI411-186 Cefepime + Amikacin Patients:

Patient AI411-186-002-006 was a 45 yo female (race unknown, from France) with acute leukemia. Recent chemotherapy (2/15/93 to ongoing during the study) included: etoposide, idarubicin, and cytarabine. On 2/23/93, she developed febrile neutropenia (Temp=38.5 C, ANC=35) and received cefepime 2gIVq12h for 21 days (2/23-3/15) and amikacin IVq12h for 20 days (2/23-3/14). Two sets of baseline blood cultures were positive for methicillin sensitive S. epidermidis (MSSE). No additional pertinent Med Hx was noted in the CRF. Pertinent concomitant therapy included: fluconazole, amphotericin B, chlormadinone, prednisone, methylergotamine, morphine, granisetron, dexchloropheniramine, lasix, zolpidem, nadroparine, PRBCs, and platelets. The patient initially defervesced; however, on Day 3, blood cultures grew both MSSE and MRSE and vancomycin was added on 2/27 (Day 5). On 3/10, amphotericin B was added due to continued fevers. On 3/14, amikacin was discontinued, and on 3/15 cefepime and amphotericin B were discontinued. On 3/15, vancomycin was changed to teicoplanin to treat a suspected catheter-related infection thought to be the source of the MRSE bacteremia. On (b) (6) the patient’s neutropenia resolved, she was afebrile, and she was discharged. On (b) (6) , the patient returned for consolidation chemotherapy and on 4/1 she developed another episode of febrile neutropenia and was started on ceftazidime and amikacin. Fever continued and ciprofloxacin and amphotericin B were added on 4/10 and 4/11, respectively. On 4/12, blood cultures grew E. cloacae and the patient developed acute respiratory distress and hemoptysis requiring ICU care. On (b) (6) days post cefepime), she died. No autopsy was performed. The investigator deemed that the patient’s death was due to E. cloacae sepsis and pneumonia secondary to profound chemotherapy-induced neutropenia and unrelated to cefepime. No SAEs were noted in the CRF.

M.O. comment: This patient’s death was most likely due to E. cloacae sepsis and pneumonia secondary to chemotherapy-induced neutropenia and leukemia.

Patient AI411-186-002-007 was a 62 yo female (race unknown, from France) with acute leukemia. Recent chemotherapy (2/17-2/23/93) included: cytarabine and mitoxantrone. On 2/25/93, she developed febrile neutropenia (Temp=39.1 C, ANC=800, [ANC=400 on 2/26]) and received cefepime 2gIVq12h and amikacin IVq12h for 8 days (2/25-3/4). One out of three baseline blood cultures was positive for E. coli susceptible to study therapy. No additional pertinent Med Hx was noted in the CRF. Pertinent concomitant therapy included: prednisone, zolpidem, paracetamol, codeine, ranitidine, clonazepam, granisetron, metoclopramide, hydrocortisone, phloroglucinol, tiemonium, morphine, nicardipine, PRBCs, platelets, and amphotericin B. No laboratory values were provided in the CRF. On 2/26, the subject was afebrile. Follow-up blood cultures through 3/3 were negative. Thrombocytopenia persisted despite platelet transfusions. On 3/3 she developed altered mentation and on 3/4 was diagnosed with severe cerebral hemorrhage and both cefepime and amikacin were discontinued on the same day. The patient died on (b) (6) days post therapy). Autopsy status was not provided in the CRF.

The investigator deemed that the patient’s death was due to cerebral hemorrhage secondary to severe thrombocytopenia due to chemotherapy and leukemia and unrelated to study therapy. No SAEs were noted in the CRF.

M.O. comment: This patient’s death was due to cerebral hemorrhage secondary to thrombocytopenia likely from chemotherapy and leukemia. Given that the CRF did not contain baseline and post baseline platelet counts, the M.O. cannot rule out the possibility that study therapy may have contributed to the patient’s thrombocytopenia, though this is a less likely etiology than chemotherapy and leukemia.

Patient AI411-186-004-006 was a 56 yo female (race unknown, from France) with acute non Hodgkin’s lymphoma who received an autologous bone marrow transplant on 2/19/93. Recent chemotherapy (2/13-2/17) included: cytarabine and melphalan; she also received radiotherapy (2/10-2/12/93). On 2/19/93, she developed febrile neutropenia (Temp=40.0 C, ANC<20) and received cefepime 2gIVq12h for 7 days (2/19-2/25) and amikacin 650 mgIVq12h for 5 days (2/19-2/23). Baseline urine culture grew E. coli and blood cultures were negative. CXR was negative on 2/18. No additional pertinent Med Hx was noted in the CRF. Pertinent concomitant therapy included: antimicrobials as discussed below, pefloxacin, fluconazole, metoclopramide, methylprednisolone, ondansetron, heparin, paracetamol, furosemide, insulin, dopamine, PRBCs, platelets, G-CSF, bromhexidine, and ganciclovir. The patient’s condition progressively worsened over the course of the hospitalization. On 2/21, she remained febrile, and began developing respiratory and renal dysfunction. Vancomycin and acyclovir were added and the patient was transferred to the ICU for mechanical ventilation. On 2/23, the patient underwent bronchoscopy. BAL culture grew respiratory syncytial virus (RSV). Amikacin was stopped on 2/23, and cefepime was changed to piperacillin/tazobactam on 2/25. Repeat urine culture on 2/29 was negative. The patient was placed on aerosolized ribavirin for RSV; however, CXRs demonstrated progressively worsening interstitial pneumonia. The patient died (b) (6) (b) on days post cefepime and(6) days post amikacin). No autopsy was performed. The investigator deemed that the patient’s death was due to respiratory failure caused by RSV pneumonia and pancytopenia following chemotherapy. The following SAEs were noted in the CRF: “RSV pneumonia” (2/22-3/26) and “acute renal insufficiency” requiring hemodialysis (2/22-3/26) were deemed unrelated to study therapy.

M.O. comment: This patient’s death was most likely due to RSV pneumonia. The M.O. cannot rule out the possibility that study therapy contributed to the patient’s renal failure.

AI411-186-004-007 was a 42 yo male (race unknown, from France) with acute lymphoblastic leukemia. Recent chemotherapy (2/13-2/19/93) included: cytarabine, dexamethasone, mitoxantrone, etoposide, and asparaginase. On 2/20/93, he developed febrile neutropenia (Temp=40.0 C, ANC <40) and received a single dose of cefepime 2gIVq12h plus amikacin 490mgIVq12h (2/20 only). One of 3 baseline blood cultures was positive for methicillin- sensitive S. aureus (MSSA). Pertinent Med Hx included: s/p autologous bone marrow transplant (BMT) on 4/29/92. Pertinent concomitant therapy included: Amphotericin B (2/17-2/19), vancomycin (2/17-2/19), amikacin 2/17-2/19, dopamine, noradrenaline, heparin, G-CSF, paracetamol, Lasix, ranitidine, acyclovir, platelets, PRBCs, fresh frozen plasma. One hour after the first doses of cefepime and amikacin, the patient developed septic shock. Study therapy was

stopped after the first dose; and teicoplanin, pefloxacin, ceftazidime, and acyclovir were started. CXR on 2/20 revealed bilateral pulmonary edema. The patient died on (b) (6) days post study therapy). No autopsy was performed. The investigator deemed that the patient’s death was caused by worsening septic shock from MSSA and was not deemed attributable to study therapy. The following was a SAE noted during the course of the study: septic shock (2/20 to death) deemed unrelated to study therapy.

M.O. comment: This patient’s death was most likely due to MSSA septic shock. The M.O. cannot rule out the possibility of study therapy failure, though the patient only received a single dose of cefepime before being switched to a multi-antibacterial regimen that included ceftazidime.

Patient AI411-186-004-021 was a 42 yo male (race unknown, from France) with non-Hodgkin’s lymphoma s/p autologous bone marrow transplant. Recent chemotherapy (5/12 to ongoing during the study) included: cytarabine, melphalan, and methylprednisolone; he also received radiotherapy (5/17-5/19). On 5/19/93, he developed febrile neutropenia (Temp=38.5 C, ANC=144) and received cefepime 2gIVq12h for 13 days (5/19-5/31) and amikacin 500mgIVq12h for 10 days (5/19-5/28). Seven baseline blood cultures grew Acinetobacter iwoffii and E. coli both susceptible to study therapy and MRSE. CXR was negative on 5/17. No additional pertinent Med Hx was noted in the CRF. Pertinent concomitant therapy included: acyclovir, chlorazepate, metoclopramide, heparin, lorazepam, flunitrazepam, omeprazole, ondansetron, calcium, methylprednisolone, G-CSF, PRBCs, platelets, paracetamol, and vitamin B12. Vancomycin was added on 5/22. CXR on 5/24 revealed interstitial pneumonia. Additional antimicrobial therapy was added as follows: ciprofloxacin (5/25), amphotericin B (5/27), bactrim and ganciclovir (5/28), and imipenem (5/31). Amikacin was stopped on 5/28 and cefepime was stopped on 5/31. Repeat blood cultures from 5/21 to 6/2 were negative. Bronchoscopy on 6/2 was negative though CXR progressively demonstrated worsening diffuse pneumonia. Head CT on 6/6 showed multiple abscesses. The patient lapsed into a coma and died with respiratory (b) (6) (b) failure on days post cefepime and (6) days post amikacin). Autopsy revealed disseminated Aspergillus fumigatus infection (in lungs and brain). The investigator deemed that the patient’s death was due to disseminated Aspergillus fumigatus infection. The following SAE was noted in the CRF: “diffuse Aspergillosis” (6/8) was deemed unrelated to study therapy.

M.O. comment: This patient’s death was most likely due to disseminated Aspergillus fumigatus infection.

Patient AI411-186-015-004 was a 58 yo male (race unknown, from France) with acute non- lymphocytic leukemia. Recent chemotherapy (6/5-6/11/93) included: cytarabine and amsacrine. On 6/16/93, he developed febrile neutropenia (Temp=39.2 C, ANC<100) and received cefepime 2gIVq12h for 2 days (6/16-6/17) and amikacin 400mgIVq12h for 17 days (6/16-7/2). One of two baseline blood cultures grew Streptococcus mitis (resistant to study therapy); urine culture was negative. CXR was negative on 6/14. No additional pertinent Med Hx was noted in the CRF. Pertinent concomitant therapy included: furosemide, heparin, alizapride, chlorpromazine, dexamethasone, vitamin K, paracetamol, metoclopramide, PRBCs, platelets, dobutamine, dopamine, G-CSF, isosorbide dinitrate, fentanyl, midazolam, and pancuronium. The patient’s status worsened within hours of enrollment. Vancomycin was added on 6/16. Cefepime was

discontinued on 6/17 after only 4 doses. The patient’s respiratory status progressively worsened culminating in ARDS requiring transfer to the ICU and intubation. Additional antimicrobial (b) agents were added: piperacillin, bactrim, ganciclovir, amphotericin B, and spiramycin. On (6) (b) (b) ( (6) days post cefepime and (6)days post amikacin), his pulmonary status worsened, he became anuric, and died. No autopsy was performed. The investigator deemed that the patient’s death was due to S. mitis sepsis leading to acute respiratory failure. The following SAE was noted in the CRF: “ARDS” (6/19-7/4) was deemed unrelated to study therapy.

M.O. comment: This patient’s death was most likely due to S. mitis sepsis leading to ARDS and multiple organ failure. The S. mitis was reported as resistant to cefepime.

Patient AI411-186-020-004 was a 45 yo female (race unknown, from France) with acute non- lymphocytic leukemia. Recent chemotherapy (4/2-4/11/93) included: cytarabine, lomustine, and doxorubicin. On 4/8/93, she developed febrile neutropenia (Temp=not provided, ANC<400) and received cefepime 2gIVq12h for 14 days (4/8-4/21/93) and amikacin 375mgIVq12h for 16 days (4/8-4/23). Baseline cultures were reportedly negative. No additional pertinent Med Hx was noted in the CRF. Pertinent concomitant therapy included: colistin, neomycin, amphotericin B, nalidixic acid, G-CSF, nicardipine, omeprazole, paracetamol, droperidol, phloroglucinol, PRBCs, and platelets. Vancomycin was added on 4/11 because of persistent fever. Over the following 2 weeks, the patient manifested signs and symptoms of hepatic failure (icterus, hepatomegaly, tense abdomen, encephalopathy, total bilirubin increased from 12 µmol/L to 318 µmol/L by 4/23). The investigator initially suspected veno-occlusive disease. Serum AST, ALT, and alkaline phosphatase reportedly remained within normal limits. On 4/18, metronidazole and amphotericin B were added for a suspected intra-abdominal infection. On 4/21, the patient was diagnosed as having life-threatening hepatic failure, severe renal impairment, severe alveolar hemorrhage, was transferred to the ICU, cefepime was discontinued, and imipenem was started. On 4/23, amikacin was discontinued due to worsening renal function. (b) (6) (b) On days post cefepime and (6) days post amikacin), the patient died in acute hepatic failure with encephalopathy. No autopsy was performed. The investigator deemed that the patient’s death was due to probable intra-abdominal sepsis (pathogen unknown) and severe aplasia secondary to underlying leukemia and chemotherapy, and unrelated to study therapy. The following SAEs were noted in the CRF: “hepatic failure” (4/21 to 4/25), “renal impairment” ((4/21 to 4/25), and “alveolar hemorrhage” (4/23-4/25). All were deemed unrelated to study therapy.

Patient AI411-186-020-011 was a 61 yo male (race unknown, from France) with acute non- lymphocytic leukemia. Recent chemotherapy (6/11/93-ongoing during study) included: cytarabine. On 6/12/93, he developed febrile neutropenia (Temp=39.0 C, ANC=300) and received cefepime 2gIVq12h for 40 days (6/12-7/21) and amikacin IVq12h for 33 days (6/12 7/14). Baseline cultures were negative. CXR was negative on 6/17. No additional pertinent Med Hx was noted in the CRF. Pertinent concomitant therapy included: colistin, neomycin,

amphotericin B, furosemide, ondansetron, paracetamol, diltiazem, PRBCs, platelets, vitamin K, dexchloropheniramine, fluconazole, and itraconazole. Pertinent laboratory values included: creatinine=115 µmol/L on 6/9 and 198 µmol/L on 6/18. On 6/14, vancomycin was added because of persistence of fever. On 6/17, the patient defervesced and remained afebrile for 2 weeks. On 6/30, the patient experienced new fevers. Fluconazole was initially added. On 7/7, the amikacin dose was decreased due to worsening renal function. On 7/12, his CXR revealed pulmonary opacities believed to be consistent with pulmonary aspergillosis and amphotericin B was added. On 7/23, the patient experienced facial paralysis and hemiplegia which were (b) (6) (b) attributed to a suspected Aspergillus brain abscess. On days post cefepime and (6) days post amikacin), the patient died. No autopsy was performed. The investigator deemed that the patient’s death was due to recalcitrant leukemia and disseminated aspergillosis. The following SAEs were noted in the CRF: “facial paralysis (cerebral Aspergillosis)” (7/23-7/27), “renal failure” (7/7-7/27), and “cholestasis” (6/18-7/27). All were deemed unrelated to study therapy.

Patient AI411-186-026-009 was a 39 yo female (race unknown, from France) with acute non- lymphocytic leukemia, s/p allogeneic bone marrow transplant on 4/15/93. Recent chemotherapy (4/9-4/13/93) included: cytarabine and melphalan; she also received radiotherapy (4/8/93). On 4/17, she developed febrile neutropenia (Temp=38.8 C, ANC<400) and received cefepime 2gIVq12h for 10 days (4/17-4/26) and amikacin 400mgIVq12h for 7 days (4/17-4/23). Baseline cultures were negative. CXR was negative on 4/17. No additional pertinent Med Hx was noted in the CRF. Pertinent concomitant therapy included: cyclosporine, methotrexate, heparin, furosemide, alizapride, ranitidine, alprostadil, morphine, hydrocortisone, dexchloropheniramine, acyclovir, PRBCs, platelets, dopamine, vitamin K, phloroglucinol, chlorazepate, prostaglandin E1, polyvalent immunoglobulin, methylprednisolone, octreotide, ofloxacin, fluconazole, imipenem, and vancomycin. Pertinent laboratory values included: creatinine=62 µmol/L on 4/17 and 226 µmol/L on 4/26; total bilirubin=17 µmol/L on 4/19 and 33 µmol/L on 4/26. During the course of the study, acyclovir and vancomycin were initially added for persistent fever though no pathogen was identified on repeat cultures. On 4/23, the patient was noted to have worsening hepatic and renal function, and amikacin was discontinued. On 4/26, cefepime was changed to imipenem due to worsening status and suspected severe sepsis (pathogen not identified). On 4/27, the patient experienced severe respiratory distress and lapsed (b) (6) (b) into coma. On days post cefepime and (6)days post amikacin), the patient died. Autopsy was reportedly performed; however results were not provided in the CRF. The investigator deemed that the patient’s death was due to multiple organ failure possibly secondary to complications of the bone marrow transplant and unrelated to study therapy. The following SAEs were noted in the CRF: “renal impairment” (4/20-4/28) and “hepatic impairment” (4/23 4/28) were deemed unrelated to study therapy.

(study therapy and/or concomitant medications). The M.O. cannot rule out the possibility that study therapy may have contributed to the patient’s hepatic and renal failure.

Patient AI411-186-026-011 was a 37 yo female (race unknown, from France) with acute lymphoblastic leukemia, s/p bone marrow transplant on 5/6/93 (Day 3 of study therapy). Recent chemotherapy (4/30-5/4/93) included: cytarabine and melphalan; she also received radiotherapy (4/26-4/29/93). On 5/4, she developed febrile neutropenia (Temp=39.8 C, ANC<200) and received cefepime 2gIVq12h for 10 days (5/4-5/13) and amikacin 500mgIVq12h for 4 days (5/4 5/7). Baseline blood cultures were negative. CXR was negative. No additional pertinent Med Hx was noted in the CRF. Pertinent concomitant therapy included: heparin, cyclosporine, alizapride, ranitidine, furosemide, ondansetron, octreotide, clobutinol, loperamide, fluoxetine, tropatepine, platelets, PRBCs, morphine, pethidine, phloroglucinol, chlorazepate, methotrexate, G-CSF, dopamine, polyvalent immunoglobulin, fluconazole, and amphotericin B. Pertinent laboratory values included: serum creatinine was 79 µmol/L on 5/3 and 315 µmol/L on 5/17. On 5/6, the patient underwent bone marrow transplant and was placed on methotrexate and cyclosporine as prophylaxis against graft-versus-host disease. IV acyclovir was also added. Fever persisted despite negative cultures. On 5/7 amikacin was discontinued due to worsening renal function and ciprofloxacin was added. On 5/8, a single blood culture grew a Streptococcus sp. (per CRF, susceptibilities were not performed because investigator considered this a “contaminant”) and vancomycin was added. On 5/10, the patient developed signs of shock and multiple organ failure, and was transferred to the ICU. On 5/13, cefepime was changed to imipenem and on 5/14 amphotericin B was added. The patient died on (b) (6) days post (b) cefepime and (6) days post amikacin). No autopsy was performed. The investigator deemed that the patient’s death was due to sepsis (pathogen unknown), multiple organ failure, and renal failure due to methotrexate and cyclosporine administration, and unrelated to study therapy. The following SAE was noted in the CRF: “renal impairment” (5/10-5/27) was deemed unrelated to study therapy.

M.O. comment: This patient’s death was due to multiple organ failure. The differential diagnosis includes: severe sepsis/septic shock due to an unknown pathogen (possibly a viridans group Streptococci, cannot rule out study therapy failure), failed bone marrow transplant with graft versus host disease, acute leukemia, and drug toxicity (study therapy and/or concomitant medications). The M.O. cannot rule out the possibility that study therapy may have contributed to the patient’s renal failure. The M.O. also cannot rule out the possibility that this patient may have developed serotonin syndrome because this patient was on a number of agents linked to this syndrome (fluoxetine, pethidine, morphine, alizapride, and tropatepine).

Patient AI411-186-028-001 was a 38 yo female (race unknown, from France) with acute myelogenous leukemia. Recent chemotherapy (12/3-12/12/92) included: daunorubicin, cytarabine, and mitoxantrone. On 12/16/92, she developed febrile neutropenia (Temp=39.1 C, ANC=0) and received cefepime 2gIVq12h and amikacin 500mgIVq12h for 20 days (12/16/92 1/4/93). Baseline blood cultures were negative. Urine culture grew E. coli which was deemed a contaminant by the investigator. CXR showed a possible left sided pneumonia on 12/15/92. No additional pertinent Med Hx was noted in the CRF. Pertinent concomitant therapy included: tobramycin, colistin, amphotericin B, cimetidine, dexchloropheniramine, furosemide,

norethisterone, PRBCs, platelets, ketoprofene, paracetamol, methylprednisolone, chlorazepate, alprazolam, G-CSF, and dopamine. Pertinent laboratory values included: percent blast cells were 2% on 12/16/92 and 72% on 1/4/93. Vancomycin was added on 12/22 due to persistence of fever. CXR on 12/23 showed bilateral pneumonia and serum serology was positive for Aspergillus sp. Amphotericin B and itraconazole were added for coverage of presumptive pulmonary Aspergillosis. The patient remained aplastic and febrile for another week and died on (b) (6) day post cefepime and amikacin). An autopsy was limited to lung biopsies which revealed leukemic blast cells. The investigator deemed that the patient’s death was due to acute leukemia resistant to chemotherapy. No SAEs were noted in the CRF.

M.O. comment: This patient’s death was most likely due to acute leukemia refractory to chemotherapy. The patient had 72% blast cells on blood smear 1 day before her death and autopsy demonstrated blast cells on lung biopsy specimens.

Patient AI411-186-028-012 was a 40 yo male (race unknown, from France) with acute myelogenous leukemia (second relapse). Recent chemotherapy (4/24-4/30/93) included: idarubicin and carboplatin. On 5/3, he developed febrile neutropenia (Temp=38.2 C, ANC=0) and received cefepime 2gIVq12h and amikacin IVq12h for 25 days (5/3-5/27). (From 5/3-5/7, the patient was erroneously given cefepime 2gIVq8h.) Baseline blood cultures were negative. CXR was negative. No additional pertinent Med Hx was noted in the CRF. Pertinent concomitant therapy included: colistin, tobramycin, and amphotericin B for gut decontamination, cimetidine, dexchloropheniramine, PRBCs, platelets, sucralfate, morphine, ketoprofene, furosemide, vancomycin, IV amphotericin B, fusidic acid, and pristinomycine. Initially, the patient rapidly defervesced until 5/12 when he spiked a fever to 38.3 C. Between 5/14 and 5/16, five blood cultures grew methicillin-resistant S. haemolyticus. Vancomycin was added on 5/17 and fusidic acid was added on 5/19. The patient remained febrile and in blast crisis over the following week. On (b) (6) , study therapy was discontinued and the patient was discharged home on (b) (6) . The patient expired on (b) (6) days post study therapy). No autopsy was performed. The investigator deemed that the patient’s death was due to refractory acute leukemia. No SAEs were noted in the CRF.

M.O. comment: This patient’s death was most likely due to acute leukemia in blast crisis.

Patient AI411-186-029-002 was a 28 yo male (race unknown, from France) with high-grade non Hodgkin’s lymphoma, s/p allogeneic bone marrow transplant on 11/19/92. Recent chemotherapy (11/13-11/24/92) included: cyclophosphamide and methotrexate; he also received radiotherapy (11/16/92). On 11/25, he developed febrile neutropenia (Temp=39.7 C, ANC<100) and received cefepime 2gIVq12h for 4 days (11/25-11/28) and amikacin 530mgIVq12h for 8 days (11/25-12/2). Baseline blood and urine cultures were negative. CXR was negative. No additional pertinent Med Hx was noted in the CRF. Pertinent concomitant therapy included: nystatin, neomycin, nifluoxazide, colistin, and gentamicin for gut decontamination, acyclovir, morphine, furosemide, chlorpromazine, tiapride, dopamine, omeprazole, PRBCs, platelets, metoclopramide, ondansetron, polyvalent immunoglobulin, dopamine, and solumedrol. Pertinent laboratory values included: AST=25 on 11/25 and 869 on 12/2; ALT=31 on 11/25 and 622 on 12/2; total bilirubin =31 µmol/L on 11/25 and 383 µmol/L on 12/2; alkaline phosphatase=79 on 11/25 and 68 on 12/2; creatinine=97 µmol/L on 11/25 and 206 on 12/2. On

11/27, the patient developed severe thrombocytopenia with extensive purpura that the investigator deemed was due to veno-occlusive disease (with hepatic failure) and was unrelated to study therapy. Vancomycin was also added on 11/27 for persistence of fever. On 11/28, the patient became hypotensive (SBP=80, T=39.2) and the cefepime was changed to ciprofloxacin. The thrombocytopenia reportedly improved with transfusions and he became afebrile by 11/30. Amikacin, vancomycin and ciprofloxacin were discontinued on 12/2. The patient died on (b) (6) (b) days post cefepime and(6) days post amikacin). Autopsy was performed; however the report was not in the CRF. The investigator deemed that the patient’s death was due to veno occlusive disease and unrelated to study therapy. The following SAEs were noted in the CRF: “veno-occlusive disease of the liver” (11/27-12/22) was deemed unrelated to study therapy; however, the M.O. noted that the cefepime was discontinued on 11/28.

M.O. comment: This patient’s death was most likely due to veno-occlusive disease associated with a failed bone marrow transplant/graft versus host disease. The M.O. cannot rule out that the initial study therapy failed; however the patient did not appear to ultimately die from the septic event. Additionally, the M.O. cannot rule out the possibility that study therapy may have contributed to the patient’s hepatic and renal failure.

Study AI411-186 (Comparator = ceftazidime & amikacin):

Patient AI411-186-004-019 was a 57 yo male (race unknown, from France) with acute non Hodgkin’s lymphoma. Recent chemotherapy (4/19-4/23) included: etoposide and cytarabine. On 5/2/93, he developed febrile neutropenia (Temp=38.6 C, ANC<200) and received ceftazidime 2gIVq8h for 6 days (5/2-5/7) and amikacin 590mgIVq12h for 20 days (5/2-5/21). Baseline blood and urine cultures were negative. CXR on 5/2 showed a small left sided pleural effusion thought to be due to lymphoma. No additional pertinent Med Hx was noted in the CRF. Pertinent concomitant therapy included: oral acyclovir, vancomycin, amikacin and amphotericin B for gut decontamination; paracetamol, codeine, amitriptyline, nicardipine, PRBCs, platelets, furosemide, G-CSF, imipenem, and ciprofloxacin. Vancomycin was added on 5/6 because of persistence of fever. On 5/7 the patient developed dyspnea and CXR showed left sided pericardial infiltrates and worsening effusions. Ceftazidime was stopped and amphotericin B and piperacillin/tazobactam were added. Imipenem and ciprofloxacin were added on 5/8. The patient continued to deteriorate and developed extensive lymphoma lesions and massive bone marrow involvement. The antimicrobials were discontinued in a serial fashion, including (b) (6) ( amikacin which was stopped on 5/21. The patient died on days post ceftazidime and b days post amikacin). No autopsy was performed. The investigator deemed that the patient’s death was due to progressive non-Hodgkin’s lymphoma. No SAEs were noted in the CRF.

M.O. comment: This patient’s death was most likely due to progressive non-Hodgkin’s lymphoma.

resistant S. epidermidis (MRSE). No additional pertinent Med Hx was noted in the CRF. Pertinent concomitant therapy included: antibacterial agents as noted below, oral tobramycin and amphotericin B for gut decontamination; fraxiparine, G-CSF, lactulose, phenobarbitol, colchicine, PRBCs, platelets, dextropropoxyphene, and paracetamol. Vancomycin was added on 12/30 due to the MRSE positive blood cultures. On 1/1/93, the patient defervesced and his ANC=1688. On 1/4, ceftazidime and amikacin were discontinued due to completion of the course of therapy. On 1/6, the patient was noted to have purulent discharge around his catheter site that grew MRSE. Pefloxacin and fosfomycin were added. Repeat blood cultures from 1/8 were negative. Over the following week, the patient’s lymphoma progressed; his clinical status deteriorated, and was in a coma by 1/14. The patient died on (b) (6) (b) (6) days post ceftazidime and amikacin). No autopsy was performed. The investigator deemed that the patient’s death was due to progression of his underlying lymphoma. The following SAE was noted in the CRF: “death” (b) (6) was deemed unrelated to study therapy.

M.O. comment: This patient’s death was most likely due to progression of underlying lymphoma.

Patient AI411-186-023-001 was a 36 yo male (race unknown, from France) with Stage IV non Hodgkin’s lymphoma, s/p autologous bone marrow transplant on 3/20/93. Recent chemotherapy (3/11-3/16/93) included: carmustine, etoposide, cytarabine, and melphalan. On 3/22/93, he developed febrile neutropenia (Temp=38.8 C, ANC<100) and received ceftazidime 2gIVq8h for 2 days (3/22-3/23) and amikacin 600mgIVq12h for 11 days (3/22-4/1). Baseline blood cultures were negative. CXR showed a right-sided pneumonia. No additional pertinent Med Hx was noted in the CRF. Pertinent concomitant therapy included: antimicrobial agents as noted below, oral fluconazole, gentamicin, and neomycin for gut decontamination; loperamide, dopamine, dobutamine, furosemide, benzodiazepines, Hordenol (for diarrhea), methylprednisolone, ranitidine, and bumetanide. Pertinent laboratory values included: creatinine=124 µmol/L on 3/23 and 288 µmol/L on 4/1, total bilirubin=16 µmol/L on 3/23 and 127 µmol/L on 4/1. On 3/23, the patient developed septic shock with cardiac failure and ARDS and was transferred to the ICU. Dopamine and dobutamine were initiated. CXR revealed bilateral pneumonia and cardiac echo revealed “signs of fluid overload”. Ceftazidime was stopped and imipenem and vancomycin were started on the same day. The dose of amikacin was increased to 750mgIVq12h. On 3/25, the patient developed renal failure requiring hemodialysis. On 3/28 the patient underwent bronchoscopy and BAL culture grew MRSA. The patient developed multiple organ failure and (b) (6) (b) died on days post ceftazidime and on therapy Day (6) of amikacin). No autopsy was performed. The investigator deemed that the patient’s death was due to MRSA pneumonia and septic shock with resultant renal failure and lactic acidosis. The following SAEs were noted in the CRF: cardiac failure (3/23-4/1), renal insufficiency 3/25-4/1, respiratory insufficiency (3/23 4/1), and lactic acidosis (4/1). All were deemed unrelated to study therapy.

100

Patient AI411-186-027-007 was a 30 yo female (race unknown, from France) with non- lymphocytic leukemia. Recent chemotherapy (4/30-5/9/93) included: mitoxantrone, cytarabine, and etoposide. On 4/29/93, she developed febrile neutropenia (Temp=39.8 C, ANC=2800; however ANC fell to 1 on 5/3) and received ceftazidime 2gIVq8h for 35 days (4/29-6/2) and amikacin 250mgIVq12h for 16 days (4/29-5/14). Baseline blood cultures were negative. CXR was negative. No additional pertinent Med Hx was noted in the CRF. Pertinent concomitant therapy included: oral vancomycin, amphotericin B, and netilmicin for gut decontamination; furosemide, urateoxydase, heparin, ondansetron, dexchloropheniramine, PRBCs, platelets, teicoplanin, paracetamol, fosfomycin, IV amphotericin B, and imipenem. The patient initially defervesced by 5/4 however; her severe aplasia remained until death. On 5/13, she had a new fever to 39.0 C. Amikacin was stopped on 5/14 and teicoplanin was started on the same day. IV amphotericin B was started on 5/16. On 5/17, blood cultures grew C. albicans. By 5/22 the patient was noted to be icteric with a total bilirubin=362 µmol/L (baseline was 11 µmol/L). On 6/2, ceftazidime was replaced with imipenem due to persistent fevers. Repeat blood cultures grew C. albicans and a methicillin-resistant coagulase positive Staphylococcus species (MRSA (b) (6) (b) ?). Fosfomycin was added on 6/5. The patient died on days post ceftazidime and (6) days post amikacin). No autopsy was performed. The investigator deemed that the patient’s death was due to C. albicans septicemia and acute leukemia. The following SAE was noted in the CRF: “Septicemia with C. albicans” (5/17-6/12) was deemed unrelated to study therapy.

M.O. comment: This patient’s death was most likely due to C. albicans sepsis, acute leukemia, and possibly MRSA sepsis. The M.O. cannot rule out that study therapy may have contributed to the patient’s hepatic failure.

Study AI411-189 Cefepime Patients:

Patient AI411-189-003-002 was a 58 yo White male from the United Kingdom with an undifferentiated abdominal carcinoma. The patient received chemotherapy on 5/7/93 (names of chemotherapy agents not provided in CRF). On 5/18/93, he developed febrile neutropenia (Temp=39.0 C, ANC=50) and received cefepime 2gIVq8h for 9 days (5/18-5/26/93). At baseline, the patient presented with diarrhea, lethargy, and cough. Baseline blood, urine, and stool cultures were negative. Pertinent Med Hx included: HTN, diverticulosis, diabetes mellitus, and gout. Pertinent concomitant therapy included: allopurinol, atenolol, metformin, paracetamol, and temazepam. The patient defervesced on Study Day 6. Cefepime was discontinued on Study Day 9 and he was discharged on (b) (6) . The patient died on (b) (6) days post therapy) after undergoing a surgery to assess the operability of the tumor. No autopsy was performed. The investigator deemed that the patient’s death was caused by progressive, undifferentiated anaplastic carcinoma and unrelated to cefepime. No SAEs were noted in the CRF.

M.O. comment: This patient’s death was most likely due progressive anaplastic carcinoma.

Patient AI411-189-005-025 was a 31 yo White male from the United Kingdom with Hodgkin’s disease (including abdominal involvement). On 10/28/93, he was enrolled for the treatment of febrile neutropenia (Temp=39.1 C, ANC=0, WBC=0.3, PLAT=14,000) post chemotherapy (agents not provided) and was started on cefepime 2 g IV q24h (10/28-10/29). Blood and urine cultures from 10/28 were positive for E. coli. Pertinent Med Hx included: “longstanding cancer”

101

(Hodgkin’s disease?) and urinary incontinence. Pertinent concomitant therapy included: metronidazole 400 mg on 10/29/93, amphotericin, G-CSF, paracetamol, diazepam, lorezapam, lactulose, temazepam, ondansetrone, pethidine, dopamine, morphine, and cyclizine. On 10/28, the patient was hypotensive and tachycardic (BP=70/40, HR=160). On 10/29, his BP=50/40, HR=160. Serious adverse event included: septic shock that was deemed unrelated to study therapy. An autopsy limited to the abdomen was performed. The report was reportedly sent to BMS; however no report was noted in the CRF. The investigator deemed that the patient’s death on (b) (6) was due to bacteremic shock and was not related to administration of cefepime.

M.O. comment: The patient’s death was most likely due to E. coli septic shock. The patient appeared to be in septic shock at baseline. The M.O. cannot rule out study drug failure; however, it is unclear if any antibacterial agent would have altered the patient’s clinical course. The patient’s underlying Hodgkin’s disease likely also played a role in his death.

Patient AI411-189-009-012 was a 69 yo White male from Switzerland with acute myeloid leukemia (AML) (originally diagnosed 3/93). He received chemotherapy (Ara-C and mitoxantrone) from 3/25-3/28. He developed febrile neutropenia on 4/8/93 (Temp=38.4 C, ANC=50) and cefepime 2 g IV q8h (4/8-4/14/93) was started on the same day. Baseline cultures were negative for a pathogen. Pertinent Med Hx included: HTN, 1st degree AV block, and myelodysplasia (since 1990). Concomitant therapy included: acyclovir (baseline to 4/14), ciprofloxacin 1 gram/day (baseline to 4/9), and penicillin V 1 million units/day (baseline to 4/9), amiodarone, furosemide, prednisone, paracetamol, and terfenadine (4/4-4/14). During the hospital course the fever persisted and the patient developed pulmonary edema, pericardial friction rub, and atrial flutter/fibrillation (4/16-death). The patient’s platelet counts were <10,000 from 4/8-4/14. The patient’s ANC’s were <1 and 1 on 4/8 and 4/13. The patient’s AML did not go into remission as evidenced by “massive peripheral blastosis.” The family reportedly elected for supportive care. On 4/14, the cefepime was discontinued and the patient was started on a ciprofloxacin 1500 mg/day per order of the treating physician. On(b) (6) the patient developed a cerebral hemorrhage, lapsed into coma, and died 1 hour later. Autopsy was not performed. Serious adverse events included atrial fibrillation and cerebral bleed, and neither was deemed related to study therapy. The investigator deemed that the patient’s death was most likely due to AML, cerebral bleeding, cardiac failure, and rapid atrial fibrillation.

M.O. comment: The patient’s death was most likely due to cerebral hemorrhage related to acute leukemia and associated chemotherapy.

Patient AI411-189-010-032 was a 46 yo White male from the Netherlands with acute myeloid leukemia (AML) (date of original diagnosis not found in CRF). The CRF did not note if the patient had recently received chemotherapy for AML. He developed febrile neutropenia on 4/6/94 (Temp=40.6 C, ANC=0, WBC=45.5 (88% immature forms), PLAT=11,000, Hgb=5.3) and cefepime 2 g IV (one dose on 4/6/94 hours before death) was started on the same day. Baseline blood cultures x 3 were positive for Staphylococcus aureus (MSSA). Pertinent Med Hx included: HTN. Concomitant therapy included: paracetamol and fluconazole. At baseline, the patient was noted to be pancytopenic due to leukemia and thrombocytopenic. Within hours of admission, the patient developed adult respiratory distress syndrome (ARDS) secondary to pulmonary hemorrhage thought secondary to disseminated intravascular coagulation (DIC).

102

Serious adverse events included: ARDS and cardiac arrest, and neither was deemed related to study therapy. The investigator deemed that the patient’s death on (b) (6) was due to underlying leukemia with pulmonary hemorrhage due to DIC and that the MSSA infection was not the primary cause of death.

Patient AI411-189-018-010 was a 74yo White male from Finland with relapsed Hodgkin’s lymphoma. The patient received chemotherapy on 4/11/95 (names of chemotherapy agents not provided in CRF). On 4/20/95, he developed febrile neutropenia (Temp=38.6, ANC=100) and received cefepime 2gIVq8h for 12 days (4/20-5/1/95). Baseline blood, urine, and stool cultures were negative. Pertinent Med Hx included: HTN. Pertinent concomitant therapy included: cefuroxime (4/13-4/20/95) as prophylaxis against infection during impending neutropenia, G CSF, tobramycin (4/28-5/1), triamterene, “trichlormethiazid”, “klonidinhydrochlorid”, “thematzepam”, ranitidine, and “metamizol”. Pertinent laboratory values: serum creatinine=155 µmol/L on 4/20, 264 µmol/L on 4/30 and 467 µmol/L on 5/2. Initially, the patient reportedly defervesced and neutrophil counts improved (on G-CSF) by 4/26/95. On 4/27 (Study Day 7), fever returned and on 4/28, tobramycin was added to the regimen. Over the next several days, serum creatinine increased, and on 5/1 cefepime was discontinued due to lack of efficacy and tobramycin was stopped due to nephrotoxic effect. The patient was transferred to the ICU due to cyanosis, rales, and lethargy. Dialysis was initiated on 5/2. The patient died on (b) (6) days post therapy). An autopsy was performed and showed bilateral pneumonia, thrombosis of the left femoral vein, “congestia gravis (brain) or petechiae”, and “morbus Hodgkins (underlying disease contributing to death)”. The investigator deemed that the patient’s death was caused by progressive Hodgkin’s disease, bilateral pneumonia, and renal insufficiency. The following SAEs were noted in the CRF: “increased creatinine” and ‘urinary complications” (5/1 to death) were deemed unrelated to cefepime and likely related to tobramycin.

M.O. comment: This patient’s death was most likely due pneumonia. Progressive Hodgkin’s disease and renal failure probably contributed to the patient’s death. The M.O. cannot rule out the possibility of cefepime study drug failure. Tobramycin therapy appeared to closely correlate with the patient’s renal failure; however, the M.O. cannot rule out the possibility that cefepime may have contributed to the patient’s renal failure.

Patient AI411-189-019-011 was a 63 yo White male from Sweden with acute lymphoblastic leukemia (with CNS involvement). Recent chemotherapy (3/14/94) noted in the CRF (names not provided). On 3/31/94, he developed febrile neutropenia (Temp=39.1 C, ANC=2) and received cefepime 2gIVq8h for 6 days (3/31-4/5/94). Baseline blood culture was positive for K. pneumoniae. No additional pertinent Med Hx was included in the CRF. Pertinent concomitant therapy included: ceftazidime (3/10-3/18/94), vancomycin (3/14-3/20/94), allopurinol, acyclovir, fluconazole, lactulose, benzodiazepines, pethidine hydrochloride, PRBCs and platelets. Repeat blood culture on 4/5/94 was negative. On 4/5 cefepime was discontinued due to skin rash and the patient was placed on ciprofloxacin 800 mg/day until 4/14/94. The patient was discharged on

103

(b) (6) days post therapy) the patient died. No autopsy was performed. The investigator deemed that the patient’s death was caused by acute lymphoblastic leukemia. The following AEs were noted in the CRF: erythematous rash (4/5/94 only) of moderate severity was deemed probably related to cefepime; death (b) (6) (SAE) was deemed unrelated to cefepime.

M.O. comment: This patient’s death was most likely due to acute lymphoblastic leukemia.

Patient AI411-189-022-001 was a 68 yo White female from Sweden with multiple myeloma. Recent chemotherapy (2/9/94) and radiotherapy (2/14-/25/94) noted in the CRF (names of chemotherapy not provided). On 2/18/94, she developed febrile neutropenia (Temp=38.7 C, ANC=270) and received cefepime 2gIVq8h for 8 days (2/18-2/25/94). Baseline blood and urine cultures were negative. No sputum cultures were noted in the CRF. Baseline CXR revealed bibasilar pneumonia. No additional pertinent Med Hx was included in the CRF. Pertinent concomitant therapy included: acyclovir, erythropoietin, pamidronate, paracetamol, morphine, and “dextropropoxiphen”. Cefepime was discontinued on 2/25/94 because fevers had apparently resolved; however, physical exam continued to reveal bibasilar rales (2/27), and fever returned on 2/28 (T=39.0 C) and 3/1 (T=38.3 C). On (b) (6) days post therapy) the patient died. No autopsy was performed. The investigator deemed that the patient’s death was caused by progressive multiple myeloma and pneumonia. The following SAE was noted in the CRF: “somnolence’ (2/24 to death) was deemed unrelated to cefepime and resulted in death.

M.O. comment: This patient’s death was most likely due to persistent pneumonia and progressive multiple myeloma. The M.O. cannot rule out the possibility of cefepime study drug failure.

Patient AI411-189-022-013 was a 73 yo White male from Sweden with chronic lymphocytic leukemia (CLL). Recent chemotherapy (9/18/94) noted in CRF (names not provided). On 9/23/94, he developed febrile neutropenia (Temp=38.5 C, ANC=200) and received cefepime 2gIVq8h for 12 days (9/23-10/3/94). Baseline blood and urine cultures were negative. Pertinent Med Hx included: atrial fibrillation and thyrotoxicosis. Pertinent concomitant therapy included: clotrimazole, fluconazole, digoxin, levothyroxine, “cortison aletat”, furosemide, amiodarone, nitroglycerin, enalapril, erythropoietin, prochlorperazine, PRBCs, and platelets. On 10/3, study therapy was discontinued because neutrophil count had improved and no source of infection was identified. The patient was discharged on (b) (6) days post therapy), the patient died of end stage CLL and heart failure. No autopsy was performed. The investigator deemed that the patient’s death was caused by end stage leukemia (CLL) and heart failure. The following SAE was noted in the CRF: “death” ( (b) (6) was considered unrelated to study therapy.

M.O. comment: This patient’s death was most likely due to end stage leukemia (CLL) and heart failure.

Study AI411-189 (Comparator = ceftazidime):

Patient AI411-189-006-014 was a 48 yo Black male from the United Kingdom with acute myelogenous leukemia (AML) and myelofibrosis. Recent chemotherapy (9/17-9/21/93)

104

included: mitoxantrone, cytarabine. On 9/18/93, he developed febrile neutropenia (Temp=39.4 C, ANC=435) and received ceftazidime 2gIVq8h for 4 days (9/18-9/21/93). Baseline blood cultures were negative. Pertinent Med Hx included: bone pain related to leukemia. Pertinent concomitant therapy included: antibacterial agents as noted below, colistin, dihydrocodeine, allopurinol, ondansetron, cotrimoxazole, tranexamic acid, platelets and PRBCs. Pertinent laboratory values: platelet count= 20,000 on 9/18 and 6,000 on 9/24 (last value recorded in CRF). On 9/20, the patient began having hematuria secondary to thrombocytopenia related to AML and chemotherapy. On 9/21, ceftazidime was discontinued due to persistent fevers and the patient was placed on imipenem, amphotericin B, and vancomycin (later changed to teicoplanin) with resolution of fever. On 9/29, he developed rectal bleeding. The patient died on (b) (6) days post therapy) due to an acute GI bleed. Autopsy was performed; however the report was not provided in CRF. The investigator deemed that the patient’s death was caused by GI bleed and AML and was unrelated to study therapy. The following SAE was noted in the CRF: bleeding from 9/20 to death (rectal, hemoptysis, hematuria, epistaxis) was considered unrelated to study therapy.

M.O. comment: This patient’s death was most likely due to GI bleed secondary to thrombocytopenia related to acute leukemia and chemotherapy. The M.O. cannot rule out study drug failure given persistent fevers; however the septic event did not appear to be the ultimate cause of the patient’s death.

Patient AI411-189-007-015 [a.k.a. AI411-189-007-010] was a 37 yo White male from the United Kingdom with high-grade T-cell lymphoma. No prior chemotherapy was noted in the CRF. On 11/15/93, he developed febrile neutropenia (Temp=39.7 C, ANC=unknown) and received ceftazidime 2gIVq8h for 2 day (11/15-11/16/93). The patient was mistakenly enrolled in the study. He had received ceftazidime three days prior to enrollment. The results of baseline blood, urine, and sputum cultures were not recorded in the CRF. No additional pertinent Med Hx was included in the CRF. Pertinent concomitant therapy included: ceftazidime (11/10 11/12/93), ciprofloxacin (11/12-11/15/93), imipenem 11/16-11/19/93), ranitidine, and Buscopar. Laboratory values were not recorded in the CRF. Once it was determined that the patient was mistakenly enrolled in the study, ceftazidime was discontinued on 11/16 and switched to imipenem. The patient underwent a liver biopsy complicated by bleeding into the peritoneum, renal failure, and ARDS. The patient died on (b) (6) days post therapy). The investigator deemed that the patient’s death was caused by relapsed lymphoma.

M.O. comment: This patient’s death was most likely due to complications related to the liver biopsy and relapsed lymphoma.

Patient AI411-189-009-041 was a 56 yo White female from Switzerland with toxic granulocytopenia/thrombocytopenia secondary to methotrexate therapy for rheumatoid arthritis. On 7/7/94, she developed febrile neutropenia (Temp=39.3 C, ANC<400) and received ceftazidime 2gIVq8h for 6 days (7/7-7/12/94). Baseline blood cultures were negative. Baseline urine and throat cultures grew C. albicans. Pertinent Med Hx included: retroperitoneal hemorrhage, pulmonary embolism, cholecystectomy, rheumatoid arthritis, septic arthritis, splenectomy, tobacco use, and steroid therapy. Pertinent concomitant therapy included: fluconazole, amoxicillin/clavulanate, norfloxacin, ceftriaxone, imipenem, clarithromycin,

105

trimethoprim/sulfamethoxazole, prednisone, oxazepam, tramadol, ranitidine, furosemide, paracetamol, metoclopramide, G-CSF, and hydrocortisone. The patient continued ceftazidime until 7/12 when she was noted to be afebrile and neutropenia had apparently resolved on G-CSF. On 7/15, the patient was again febrile and was treated with amoxicillin/clavulanate, norfloxacin, and ceftriaxone. She continued to deteriorate. On 7/18, BAL revealed P. carinii pneumonia and therapy was changed to imipenem, clarithromycin, and trimethoprim/sulfamethoxazole. She developed a DVT and received heparin therapy. On 7/26, the patient reportedly developed hemorrhagic shock refractory to pressors and died on (b) (6) Autopsy was performed and showed pericarditis with pericardial effusion. The investigator deemed that the patient’s death was caused by hemorrhagic shock with pericardial tamponade secondary to thrombocytopenia and heparin therapy and unrelated to ceftazidime. The following SAEs were noted in the CRF: PCP pneumonia (7/18 to death) and hemorrhagic shock (7/26 to death) were deemed unrelated to study therapy.

M.O. comment: This patient’s death was most likely due to hemorrhagic shock and pericardial tamponade. PCP pneumonia likely also contributed to the patient’s death.

Patient AI411-189-009-044 was a 31 yo White female from Switzerland with Non-Hodgkin’s lymphoma. Recent chemotherapy (8/10/94) included: Ara-C and VP-16. On 8/28/94, she developed febrile neutropenia (Temp=38.5 C, ANC<200) and received ceftazidime 2gIVq8h for 17 days (8/28-9/13/94). Baseline cultures were negative. Pertinent Med Hx included: Non Hodgkin’s lymphoma originally diagnosed in 1989, epigastric pain, and mucositis. Pertinent concomitant therapy included: ciprofloxacin (8/27-8/28), penicillin V (8/27-8/28), acyclovir, fluconazole, vancomycin (9/8-9/13), netilmicin (9/8-9/12), amoxicillin (9/13-9/18), ciprofloxacin (9/13-9/19), piperacillin/tazobactam (9/19-9/21), sucralfate, omeprazole, G-CSF, tramadol, paracetamol, benzodiazepines, furosemide, ondansetron, “MUTHESA (R)-CH”, and progesterone. Pertinent laboratory values: AST=20 on 8/27 and 865 on 9/20, ALT=13 on 8/27 and 1420 on 9/20, creatinine=72 µmol/L on 8/27 and 194 µmol/L on 9/20. The patient initially defervesced on Day #4; however, on 9/7, she became dsypneic and again febrile (39.5 C). Repeat blood and urine cultures were positive for Enterococcus sp. (resistant to ceftazidime and intermediate to vancomycin). Ceftazidime was discontinued on 9/7. A variety of antibacterial agents were prescribed in an effort to treat the Enterococcal bacteremia and UTI. CXR showed pleural effusions and cardiomegaly. Dilated cardiomyopathy was diagnosed on 9/7 by echocardiogram and was thought to be due to chemotherapy. The patient continued to deteriorate and went into multisystem organ failure (cardiogenic shock, renal and hepatic failure) on 9/20. The patient died on (b) (6) days post therapy). Autopsy was performed; however the report was written in Swiss. (The report appeared to note infiltrative lymphoma, dilated cardiomyopathy, pericardial effusion, pulmonary edema with bilateral pleural effusions, acute cholecystitis, and generalized icterus.) The investigator deemed that the patient’s death was caused by dilated cardiomyopathy (due to chemotherapy) resulting in multiple organ failure (renal and hepatic), Enterococcal bacteremia and UTI; and was unrelated to study therapy. The following SAEs were noted in the CRF: dyspnea (9/7 to death), Enterococcal UTI and bacteremia (9/7 to death), acute cholecystitis (9/? to death), cardiogenic shock with hepatic and renal failure (9/18 to death). None of these SAEs were considered related to study therapy.

106

M.O. comment: This patient’s death was most likely due to cardiogenic shock with multiple organ failure (renal and hepatic) plus Enterococcal bacteremia and UTI. The Enterococcal infections would be resistant to ceftazidime. The M.O. cannot rule out the possibility that ceftazidime may have contributed to the hepatic and renal failure.

Patient AI411-189-010-026 was a 61 yo White male from the Netherlands with Non-Hodgkin’s lymphoma. Recent chemotherapy (12/31/94) was noted in the CRF (names not provided). On 1/11/94, he developed febrile neutropenia (Temp=39.4 C, ANC=120) and received ceftazidime 2gIVq8h for 7 days (1/11-1/17/94). Baseline urine culture grew E. coli. Blood cultures were negative. Pertinent Med Hx included: HTN and urolithiasis. Pertinent concomitant therapy included: amphotericin B, temazepam, valium, lactulose, acyclovir, furosemide, albumin, and acebutolol chloride. The patient defervesced on therapy and repeat urine culture was negative on 1/14. On 1/17, ceftazidime was discontinued and on(b) (6) the patient was discharged. Two weeks later, the patient was readmitted with progressive lymphoma including hepatic involvement. The patient died on (b) (6) days post therapy). An autopsy was performed and showed “Non-Hodgkin’s lymphoma particular abdominal involvement and liver infiltration causing lactic acid acidosis with shock”. The investigator deemed that the patient’s death was caused by progressive lymphoma. No SAEs were noted in the CRF.

M.O. comment: This patient’s death was most likely due to progressive lymphoma.

Patient AI411-189-010-027 was a 43 yo White male from the Netherlands with acute myeloid leukemia (AML) with myelodysplastic syndrome. Recent chemotherapy (2/12/94) was noted in the CRF (names not provided). On 2/13/94, he developed febrile neutropenia (Temp=39.1 C, ANC=480) and received ceftazidime 2gIVq8h for 12 days (2/13-2/24/94). Baseline blood, urine, and sputum cultures were negative. Pertinent Med Hx included: status post allogeneic bone marrow transplant 7 months earlier, herpes keratitis and positive tobacco use. Pertinent concomitant therapy included: cotrimoxazole, ciprofloxacin (2/7-3/3/94), fluconazole (2/7 3/3/94), amphotericin B (2/7-3/3/94), allopurinol, electrolyte supplementation, paracetamol, morphine, pethidine, temazepam, furosemide, albumin, and vitamin K. Pertinent laboratory values included: total bilirubin=9 (units unknown) on 2/13 and 137 on 2/25; creatinine=82 on 2/13 and 65 on 2/25. Over the following two weeks, the patient developed signs and symptoms of a pulmonary process. Sputum cultures on 2/19 and 2/24 grew C. albicans. CXR between 2/17 and 2/25 progressed from trace bilateral pleural effusions to diffuse bilateral infiltrates. On 2/24, ceftazidime was discontinued due to pulmonary disease progression and other antimicrobials (discussed above) were continued. The patient’s respiratory status continued to deteriorate and he died on (b) (6) days post therapy). An autopsy was performed and the CRF noted: “This patient died due to respiratory insufficiency based on ARDS, pleuritis, and edema.” The investigator deemed that the patient’s death was caused by fungal (C. albicans) pneumonia complicated by pancytopenia due to chemotherapy for leukemia. The following was a notable AE in the CRF: “progressive icterus” (2/21 to death) of moderate severity was deemed to have an unknown relationship to study drug.

M.O. comment: This patient’s death was most likely due to an acute pulmonary process leading to ARDS. It is unclear if this acute process was an infection (bacterial vs. fungal vs. other opportunistic pathogen), a pulmonary embolism, congestive heart failure, multi-organ

107

failure including hepatic and renal, or ARDS associated with chemotherapy for the acute leukemia. Given the differential diagnosis, the M.O. cannot rule out the possibility of study drug failure. The M.O. cannot rule out the possibility that study drug may have played a role in the patient’s hepatic and renal failure.

Patient AI411-189-016-005 was a 32 yo White male from the Netherlands with acute myeloid leukemia (AML) with myelodysplastic syndrome. Recent chemotherapy (9/27/93) was noted in the CRF (names not provided). On 10/11/93, he developed febrile neutropenia (Temp=39.2 C, ANC=0) and received ceftazidime 2gIVq8h for 6 days (10/11-10/16). Baseline blood and sputum cultures were negative. CXR on 10/11 was negative. No additional pertinent Med Hx was noted in the CRF. Pertinent concomitant therapy included: antibacterial agents as discussed below, norfloxacin (9/24-10/19), amphotericin B (10/21-10/26), itraconazole, potassium, and loperamide. By Study Day #3 (10/13), the patient had developed bilateral airspace disease on CXR. BAL culture yielded Micrococcus sedentarius resistant to study therapy. On 10/14, vancomycin and co-trimoxazole were added. On 10/15, he developed severe respiratory insufficiency and was intubated. Ceftazidime was stopped on 10/16. The patient progressed to ARDS and died on (b) (6) days post therapy). An autopsy was performed and revealed: “…lungs weighed 5 kilo and were diffusely infiltrated; in the left upper lobe an abscess was found with a diameter of 7 cm. Further findings: 1500ml ascites and a septic spleen.” The investigator deemed that the patient’s death was caused by AML and neutropenia post chemotherapy and ARDS due to Micrococcus sedentarius. The following SAEs were noted in the CRF: ‘respiratory insufficiency’ (10/13 to death) and ‘progressive respiratory failure resulting in death’ (b) (6) . Neither was considered related to study therapy.

M.O. comment: This patient’s death was most likely due to ARDS resulting from pneumonia from an opportunistic, study drug-resistant pathogen and due to the patient’s prolonged neutropenia post chemotherapy for acute leukemia.

Patient AI411-189-016-011 was a 65 yo White female from the Netherlands with Non Hodgkin’s lymphoma. Recent chemotherapy (12/30/93) was noted in the CRF (names not provided). On 1/4/94, she developed febrile neutropenia (Temp=39.9 C, ANC=306) and received ceftazidime 2gIVq8h for 7 days (1/4-1/10/94). Baseline blood, urine, and sputum cultures were negative. Baseline CXR was negative. No additional pertinent Med Hx was included in the CRF. Pertinent concomitant therapy included: amoxicillin (12/31-1/4), fluconazole, temazepam, paracetamol, G-CSF, and promethazine. Following enrollment, the patient remained febrile and CXR showed evidence of bilateral pneumonia. On 1/8 (Study Day 5), gentamicin (1/8-1/11) was added. On 1/10 (Study Day 7), ceftazidime was discontinued due to poor respiratory status, and the patient’s therapy was changed to imipenem (1/11-1/20) and erythromycin (1/11-1/20). On 1/11, BAL culture grew C. albicans; however, this was judged to be a contaminant. On 1/14, she was intubated. Fever eventually resolved; however pulmonary status worsened and she developed ARDS. On 1/30, the patient became febrile again and subsequently died on (b) (6) days post therapy). An autopsy was performed and revealed massive infiltrates in both lungs and cultures from the lungs grew Herpes simplex virus (HSV). The investigator deemed that the patient’s death was caused by HSV pneumonia and non Hodgkin’s lymphoma. The following SAE was noted in the CRF: ‘respiratory failure’ (1/14 to death) was deemed unrelated to study therapy.

108

M.O. comment: This patient’s death was most likely due to HSV pneumonia.

Patient AI411-189-022-003 was a 65 yo White male from Sweden with multiple myeloma. Recent chemotherapy (3/8/94) was noted in the CRF (names not provided). On 3/21/94, he developed febrile neutropenia (Temp=38.7 C, ANC=400) and received ceftazidime 2gIVq8h for 9 days (3/21-3/29/94). Baseline blood, urine, and sputum cultures were negative. CXR was negative. No additional pertinent Med Hx was included in the CRF, but based on medications probably had CAD and diabetes. Pertinent concomitant therapy included: antimicrobial agents as discussed below, famotidine, glipizide, allopurinol, nitroglycerin, furosemide, and morphine. Vancomycin (3/25-3/29) was added on Study Day 5 due to persistent fevers. On 3/29, ceftazidime and vancomycin were discontinued and imipenem (3/30-4/5) and fluconazole (3/30 4/5) were added and the patient defervesced 2 days later. On (b) (6) days post therapy), the patient died due to progressive multiple myeloma. No autopsy was performed. The investigator deemed that the patient’s death was caused by uremia secondary to progressive multiple myeloma. The following SAE was noted in the CRF: ‘terminal uremia due to progressive myeloma’ (3/30 to death) was deemed unrelated to study therapy.

M.O. comment: This patient’s death was most likely due to renal failure secondary to progressive multiple myeloma. Cannot exclude that study therapy may have played a role in the patient’s progressive renal failure; however, the underlying multiple myeloma likely played the dominant role in the renal failure.

Patient AI411-189-023-001 was a 64 yo White male from Scandinavia with acute myelogenous leukemia (AML) in blast crisis secondary to myeloproliferative disease. Recent chemotherapy (9/3/94) was noted in the CRF: high dose Ara-C and cerubidin. On 3/13/94, he developed febrile neutropenia (Temp=39.1 C, ANC=0) and received ceftazidime 2gIVq8h for < 2 days (3/13 3/14/94). Baseline blood, sputum, and urine cultures were negative. Baseline blood pressure=150/90—only BP recorded in CRF. Pertinent Med Hx included: diabetes, prostatic hypertrophy, and osteomyelitis in 1985. Pertinent concomitant therapy included: a cephalosporin (3/7-3/10/94), PRBCs, platelets, insulin, allopurinol, acyclovir, nystatin, and ciprofloxacin (3/9-3/14/94). Pertinent laboratory values included: WBC and ANC=0 on 3/13 and 3/14. The patient deteriorated rapidly within the first 24-48 hours and died after only receiving 4 doses of study therapy on (b) (6) No autopsy was performed. The investigator deemed that the patient’s death was caused by acute leukemia with sepsis (pathogen unknown) and hepatic insufficiency related to chemotherapy. The following SAE was noted in the CRF: death (b) (6) .

M.O. comment: There was not sufficient information in the CRF to determine the cause of the patient’s death. Differential diagnosis includes: severe sepsis due to an unknown pathogen secondary to profound neutropenia, leukemia in blast crisis, hepatic failure related to chemotherapy, and anaphylactic reaction to study medication (less likely given that the patient was apparently on another cephalosporin from 3/7-3/10). The M.O. cannot rule out study drug failure, though additional factors may have contributed to the patient's death.

Study AI411-198 Cefepime + Vancomycin Patients:

109

Patient AI411-198-002-018 [a.k.a. AI411-198-002-012] was a 50 yo White male from Belgium with acute myeloblastic leukemia. The patient had received recent chemotherapy (6/9/93); however the names of the agents were not included in the CRF. On 6/18/93, he developed febrile neutropenia (Temp=39.5 C, ANC=0) and received cefepime 2gIVq8h and vancomycin 750-2000 mg/day for 4 days (6/18-6/21). Baseline blood cultures grew Streptococcus sanguis (susceptible to cefepime and vancomycin) and Staphylococcus hemolyticus (only susceptible to vancomycin). Urine culture was negative. CXR was negative. No additional pertinent Med Hx was included in the CRF. Pertinent concomitant therapy included: ciprofloxacin, fluconazole, nystatin, ondansetron, cimetidine, and Dalacin. On 6/20, the patient developed dyspnea and (b) (6) (b tachypnea and Dalacin was added. On the patient developed ARDS and died (on Day) of cefepime). No autopsy was performed. The investigator deemed that the patient’s death was due to acute myeloblastic leukemia. The following SAEs were noted in the CRF: “dyspnea/tachypnea” (6/20-6/21) and “ARDS” (6/21) were deemed unrelated to cefepime.

M.O. comment: This patient’s death was most likely due to Streptococcus sanguis sepsis (complicated by ARDS) and acute leukemia. The M.O. cannot rule out study drug failure. Staphylococcus hemolyticus was only found in one blood culture; therefore it may have been a colonizer.

Patient AI411-198-002-026 was a 69 yo White female from Belgium with acute myeloid leukemia. The patient received chemotherapy (starting after enrollment, 8/3/93); however the names of the agents were not included in the CRF. On 7/27, she developed febrile neutropenia (Temp=39.3 C, ANC=1166, WBC=58,300) and received cefepime 2gIVq8h and vancomycin 1000-1500 mg/day for 23 days (7/27-8/18/93). Baseline blood and urine cultures were negative. CXR was negative. No additional pertinent Med Hx was included in the CRF. Pertinent concomitant therapy included: erythromycin, ciprofloxacin, fluconazole, nystatin, phenoxymethyl penicillin, cimetidine, levothyroxine, and allopurinol. Fever resolved on therapy; however a new fever was noted on 8/16. On 8/18, the patient was diagnosed with pneumonitis (culture negative) and was started on erythromycin. The patient was discharged to home hospice and died on (b) (6) days post cefepime). No autopsy was performed. The investigator deemed that the patient’s death was due to refractory leukemia and pneumonitis. The following SAE was noted in the CRF: “death” (b) (6) was deemed unrelated to study therapy.

M.O. comment: This patient’s death was most likely due to refractory leukemia and culture- negative pneumonitis.

110

9/26-9/28 were negative. On 9/26, the patient developed dyspnea and was found to have pneumonia on CXR. The patient progressed to ARDS, septic shock and multiple organ failure. (b) (6) (b) The patient died on (Day (6) of study therapy). Autopsy was performed; however results were not provided in the CRF. The investigator deemed that the patient’s death was due to pneumonia, ARDS, and septic shock. The following SAE was noted in the CRF: “ARDS” (9/26 9/30) was deemed to have an unknown relationship with study therapy.

Patient AI411-198-004-013 was a 68 yo White female from Belgium with acute myeloid leukemia. Recent chemotherapy (5/14-5/21/93) included: daunorubicin and Ara-C. On 5/19, she developed febrile neutropenia (Temp=38.6 C, ANC<100) and received cefepime 2gIVq8h and vancomycin 1000-2000 mg/day for 10 days (5/19-5/28). Baseline blood and urine cultures were negative. CXR was negative. Pertinent Med Hx included: arrhythmia, anal fistula, and depression. Pertinent concomitant therapy included: fluconazole, norfloxacin, ranitidine, amikacin, disopyramide, allopurinol, ondansetron, and methylprednisolone. Pertinent laboratory values included: platelet count=12,000 on 5/19 and 3,000 on 5/28. Amikacin was added on 5/27. (b) (6) (b) On (Day (6) of study therapy), the patient fell and developed a cerebral hemorrhage secondary to thrombocytopenia (platelets=3,000) and died. No autopsy was performed. The investigator deemed that the patient’s death was due to cerebral hemorrhage. The following SAE was noted in the CRF: “death” (b) (6) was deemed unrelated to cefepime.

M.O. comment: This patient’s death was due to cerebral hemorrhage secondary to thrombocytopenia from chemotherapy and leukemia. The M.O. cannot rule out that cefepime may have contributed to the patient’s thrombocytopenia.

Patient AI411-198-004-029 was a 49 yo White male from Belgium with acute lymphoblastic leukemia s/p stem cell transplant 1/22/94. The patient received chemotherapy (1/17/94); however the names of the agents were not included in the CRF. He had also received radiotherapy (1/22/94). On 1/22/94, he developed febrile neutropenia (Temp=38.0 C, ANC=0) and received cefepime 2gIVq8h for 13 days (1/22-2/3) and vancomycin 1000-2000 mg/day for 11 days (1/24-2/3). Baseline blood and urine cultures were negative. CXR was negative. Pertinent Med Hx included: hepatic steatosis, alcohol abuse, dental caries, and smoking 1 ppd. Pertinent concomitant therapy included: ciprofloxacin, fluconazole, ondansetron, ranitidine, morphine, amphotericin B, acyclovir, penicillin G, metronidazole, erythromycin, PRBCs, G CSF, Xanax, pethidine, loperamide, and electrolyte supplementation. On 1/24, vancomycin was added. On 2/1, amphotericin B was added for a suspected fungal infection. On 2/2, CXR showed a new right-sided pulmonary infiltrate. On 2/3, acyclovir was started for oral HSV. On the same day, cefepime and vancomycin were discontinued because the infection/pulmonary infiltrate was thought to be fungal in nature. On 2/4, 2/6, and 2/7, blood cultures were positive for Streptococcus oralis (sensitive to cefepime and vancomycin) and the patient was started on

111

penicillin G, metronidazole, and erythromycin. The patient died on (b) (6) days post cefepime). Autopsy was performed and the investigator noted under ‘Adverse Events’ in the CRF that “…all biopsies (were) positive for S. oralis and S. epidermidis.” The investigator deemed that the patient’s death was due to S. oralis sepsis, ARDS, and acute leukemia. The following SAEs were noted in the CRF: “Pneumopathy right lung” (2/1-2/21), “Septicemia Streptococcus oralis” (2/4-2/21), and “Death” (b) (6) ) were deemed unrelated to cefepime.

M.O. comment: This patient’s death was most likely due to Streptococcus oralis pneumonia/sepsis in this patient with mucositis and poor dentition. This appeared to be a breakthrough infection (cefepime and vancomycin stopped 1 day prior to positive blood cultures) that persisted for 18 days post therapy likely related to concomitant immune suppression due to his underlying acute leukemia. Therefore, cannot rule out study drug failure, though additional factors (severe immune suppression related to acute leukemia) may have contributed to the patient’s death.

Study AI411-198 (Comparator = ceftazidime + vancomycin):

Patient AI411-198-001-003 was a 75 yo White male from Belgium with hairy cell leukemia. Recent chemotherapy (3/26-4/1/93) included: 2-CDA. On 3/25/93, he developed febrile neutropenia (Temp=38.4 C, ANC=960) and received ceftazidime 2gIVq8h for 32 days (3/25 4/25, including 4 days of “commercially available” ceftazidime) and vancomycin IVq8h (total daily dose ranged from 500 to 2,250 mg) for 24 days (3/25-4/17). Baseline blood culture grew Staphylococcus aureus (methicillin-sensitive) and coagulase negative staphylococci. CXR was negative. Pertinent Med Hx included: scoliosis. Pertinent concomitant therapy included: “commercially available” ceftazidime (4/22-4/25), clarithromycin (4/2-4/9), amphotericin B (4/6-4/20), erythromycin (3/31-4/2), sulfamethoxazole (4/16-4/21), ofloxacin, morphine, itraconazole, furosemide, G-CSF, paracetamol, triamterene, and folic acid. Pertinent laboratory values included: creatinine=0.85 on 3/24 and 2.45 on 4/19. The patient received 28 days of ceftazidime study therapy followed by 4 additional days of “commercially available” ceftazidime. On 4/6, a CXR showed an infiltrate consistent with pneumonia; however, blood cultures were negative. There is very little information in the CRF for the patient’s course between 4/6 and his death on 4/26. The patient reportedly remained febrile through 4/17. On 4/17, vancomycin was discontinued. On 4/25, ceftazidime was discontinued. On (b) (6) (b) post ceftazidime, (6) days post vancomycin), the patient died. No autopsy was performed. The investigator deemed that the patient’s death was due to renal insufficiency (attributed to vancomycin), bilateral pneumonia, and underlying leukemia. The following three SAEs were noted in the CRF: “death” (b) (6) ) and “bilateral pneumonia” (4/23-4/26) were deemed unrelated to study therapy; and “renal insufficiency” (4/19-4/26) was deemed related to study therapy.

M.O. comment: The CRF contained little information about the patient around the time of his death. This patient’s death was most likely due to bilateral pneumonia, renal insufficiency, and acute leukemia. The M.O. cannot rule out study therapy failure or that study therapy may have contributed to the patient’s renal dysfunction.

Patient AI411-198-001-014 was a 60 yo White male from Belgium with non-Hodgkin’s lymphoma. The patient received chemotherapy (4/18/93); however the names of the agents were

112

not included in the CRF. On 4/27/93, he developed febrile neutropenia (Temp=39.6 C, ANC<100) and received ceftazidime 2gIVq8h for 8 days (4/27-5/4) and vancomycin 800mgIVq8h for 7 days (4/27-5/3). Baseline blood and urine cultures were negative. CXR demonstrated bilateral pulmonary infiltrates. No additional pertinent Med Hx was included in the CRF. Pertinent concomitant therapy included: ofloxacin, platelets, PRBCs, ranitidine, itraconazole, G-CSF, folic acid, dobutamine, dopamine, allopurinol, furosemide, and digoxin. Pertinent laboratory values included: creatinine=0.94 (? units) on 4/27 and 1.55 (? units) on 5/4. During the hospital course, the patient developed septic shock and required mechanical ventilation. On 5/4 he was started on dialysis for worsening renal function. On(b) (6) post (b) ceftazidime and (6) days post vancomycin), the patient died. Autopsy revealed non-Hodgkin’s lymphoma in the bone marrow, bronchopneumonia, renal tubular necrosis, hepatic steatosis and “bilirubinostasis”, and myocardial hypertrophy. The investigator deemed that the patient’s death was due to septic shock and multiple organ failure. The following SAEs were noted in the CRF: “intractable septic shock” (4/27-5/5) was deemed unrelated to study therapy.

M.O. comment: This patient’s death was most likely due to septic shock and multiple organ failure. No pathogen was identified, but the focus appeared to be pneumonia. The M.O. cannot rule out study therapy failure. The patient also had active lymphoma in his bone marrow.

Patient AI411-198-002-005 was a 61 yo White female from Belgium with non-Hodgkin’s lymphoma, s/p bone marrow transplant (date of procedure not provided). The patient received chemotherapy (3/12/93); however the names of the agents were not included in the CRF. On 3/18/93, she developed febrile neutropenia (Temp=38.7 C, ANC<100) and received ceftazidime 2gIVq8h for 5 days and 2gIVq12h for 5 days (3/18-3/27) and vancomycin 850-1700mgIV total dose/day (3/18-3/20). Baseline blood and urine cultures were negative. CXR was negative. No additional pertinent Med Hx was included in the CRF. Pertinent concomitant therapy included: erythromycin, norfloxacin, itraconazole, nystatin, heparin, and G-CSF. Pertinent laboratory values included: creatinine=1.06 (? units) on 3/17 and 2.59 (? units) on 3/27. Erythromycin was added on 3/23/93 for an unknown reason. On 3/25, the patient was noted to be in respiratory distress and required intubation. CXR on the same day demonstrated (b) (6) (b) (b) pulmonary infiltrates. The patient died on (Day (6) of cefepime and (6) days post last dose of vancomycin). No autopsy was performed. The investigator deemed that the patient’s death was due to pneumonitis, ARDS, and non-Hodgkin’s lymphoma. The following SAE was noted in the CRF: “shock” (3/25-3/27) was deemed unrelated to study therapy.

M.O. comment: This patient’s death was most likely due to pneumonitis, ARDS/shock, and non-Hodgkin’s lymphoma. The M.O. cannot rule out study therapy failure, although it is unclear if the pulmonary process and shock were due to a bacterial infection or related to lymphoma or drug toxicity (chemotherapy?) because no pathogen was isolated.

Study AI411-204 Cefepime Patients: Patient AI411-204-003-042 was a 65 yo White female from the USA with lung cancer since 1992. The patient received radiation therapy and chemotherapy with VP-16 and cisplatin (dates not provided). She was hospitalized on (b) (6) for fever (39 C), neutropenia (ANC=198), right middle lobe pneumonia, and suspected septicemia. Blood cultures from 3/18/93 were positive

113

for E. coli and the patient was enrolled on cefepime 2gIVq24h (due to renal failure, creatinine=5.4 mg/dL) (3/20-3/21). Pertinent Med Hx included: emphysema, enteritis, hysterectomy, metastatic skin lesions, and peripheral edema. Pertinent concomitant therapy included: Timentin (3/19), tobramycin (3/19), albuterol, Reglan, Ativan, theophylline, Imodium, Decadron, lasix, morphine, G-CSF. On (b) (6) , the subject developed respiratory distress, hypotension, and cardiopulmonary arrest and died on the same day. Serious adverse events included: hypotension and respiratory failure deemed as having an unknown relationship with study drug. No autopsy was performed. The investigator attributed the patient’s death to lung cancer and E. coli septic shock and deemed that her death was unrelated to cefepime.

M.O. comment: The patient’s death was most likely due to E. coli septic shock. The M.O. cannot rule out study drug failure; however, the M.O. also noted that according to the CRF, the patient had received 3 doses of Timentin prior to receiving cefepime. Therefore, it is unclear if any antibacterial agent would have altered the patient’s clinical course. The patient’s underlying metastatic lung cancer likely also played a role in her death.

Patient AI411-204-003-341 was a 57 yo White male from the USA with multiple myeloma. The patient received chemotherapy with Interferon, Cytoxan, and Novantrone (dates not provided). He was hospitalized with fever and mild neutropenia (Temp=100.8 F, ANC=620) on (b) (6) Baseline blood cultures were positive for methicillin-resistant S. epidermidis (MRSE). The patient was on cefepime 2 g IV q8h (1/20-1/24/93). Pertinent Med Hx included: heart murmur, gastric bleed (7/92), chronic prosthetic knee and humeral rod osteomyelitis, depression, and penicillin allergy. Pertinent concomitant therapy included: vancomycin (1/19-1/28 and 2/1-?), Septra (1/18-1/19), Rifampin (1/18-1/19), Lanoxin, G-CSF, Zoloft, verapamil, ranitidine, MS contin, ceftazidime (2/1-2/3). During the hospital course, the patient’s fever initially resolved on vancomycin; however, by 1/24, he was noted to have purulence around the infusion catheter site on the left side of his chest. On 2/2, the patient developed a new fever to 38.8 C. The patient died on (b) (6) days post therapy) after withdrawal of life support post developing pulmonary edema, aspiration pneumonia, and new fever. The investigator deemed that the patient’s death was due to pulmonary edema, aspiration pneumonia, and multiple myeloma. Serious adverse event included: respiratory failure deemed unrelated to study drug.

M.O. comment: The patient’s death was most likely due to aspiration pneumonia and pulmonary edema likely associated with the patient’s underlying multiple myeloma. The patient’s underlying infection appeared to be due to MRSE. Potential sources of infection included the chronic prosthetic osteomyelitis and the infected catheter site. It is less likely that the patient’s death was related to study drug failure given that the MRSE was not sensitive to cefepime. It is also possible that the patient’s respiratory failure was initiated by a pulmonary embolism.

Patient AI411-204-005-142 was a 56 yo White female from the USA with metastatic adenocarcinoma of the left breast originally diagnosed in 1992 and unresponsive to chemotherapy. The patient received chemotherapy from 7/30-8/2/94 with doxorubicin, vinblastine, thiotepa, and halotestin. On 8/9/94, the patient was enrolled with febrile neutropenia (Temp=38.5 C, ANC=400, PLAT=22K, CREAT=0.6) and was placed on cefepime 2 g IV q8h for 8 days (8/9-8/16). Baseline blood and urine cultures were negative. Baseline physical exam

114

revealed 4+ lower extremity edema, oral mucosal and rectal bleeding, and tachycardia (HR=118). Pertinent Med Hx included: varicose veins, dyspnea on exertion, refractory thrombocytopenia, pancytopenia due to chemotherapy, depression, and anxiety. Pertinent concomitant therapy included: Benadryl, acetaminophen, oxycodone + acetaminophen, Lasix, Fentanyl patch, amitriptyline, prochlorperazone, alprazolam, nizatidine, vitamin K, aminocaproic acid, calcium carbonate, and potassium chloride. The patient also received platelets and PRBCs. The patient defervesced on 8/12/94 (Study Day 4) and was considered a treatment success. In the post-treatment period, she developed ileus and dyspnea. The patient died on (b) (6) days post therapy). The investigator deemed that her death was most likely due to complications related to metastatic breast cancer and refractory thrombocytopenia and unrelated to cefepime. Serious adverse event included: GI bleeding deemed unrelated to cefepime.

M.O. comment: The M.O. notes that BMS did not provide a specific etiology for the patient’s death other than “complications associated with metastatic breast cancer.” The M.O. believes that the patient’s death was most likely due to the sequelae of metastatic breast cancer. Given (b) that the patient died (6) days post therapy and no pathogen was identified during the study, it is less likely that a lack of cefepime efficacy contributed to the patient’s death. Though much less likely, and given the lack of a specific etiology for the patient’s death, the M.O. cannot completely rule out a toxic effect due to cefepime.

Patient AI411-204-005-335 was a 53 yo Black male from the USA with chronic myelogenous leukemia (CML) in blast crisis and carcinomatous meningitis. He received maintenance chemotherapy with hydroxyurea from 4/24-4/25/93 followed by induction chemotherapy with methotrexate, daunorubicin, and vincristine from 4/30-5/29/93. On 5/23/93, the patient was enrolled for treatment of febrile neutropenia (Temp=38.4 C, ANC < 100) and the patient was treated with cefepime 2 g IV q8h for 25 days (5/23-6/16/93). Blood cultures x 2 were positive for P. aeruginosa. Pertinent Med Hx included: diabetes mellitus, subclavian thrombosis requiring continuous anticoagulation, depression, and a 60-90 pack/yr smoking history. Pertinent concomitant therapy included: coumadin, amitriptyline, prednisone 40 mg daily (5/12 to unknown date), glipizide, acetaminophen, diphenhydramine, chloral hydrate, temazepam, meperidine, G-CSF, potassium, PRBCs and platelets. The patient initially defervesced on Day 2 (5/24/93), however, fever recurred on the day that cefepime was discontinued (b) (6) , ciprofloxacin 500 mg bid was started and the patient was discharged from the hospital. Of note, repeat blood cultures x2 were negative from 6/16. No serious adverse events were reported. The patient died on (b) (6) days post therapy). No autopsy was performed per the CRF. The investigator deemed that the death was due to persistent neutropenia and chronic myelogenous leukemia unresponsive to chemotherapy.

M.O. comment: The M.O. notes that BMS did not provide a specific etiology for the patient’s death other than “persistent neutropenia” and “unresponsive CML”. The M.O. believes that the patient’s death was most likely due to the sequelae of recalcitrant leukemia in blast crisis. Given that the patient died 21 days post therapy and no pathogen was identified when the fever recurred, it is less likely that a lack of cefepime efficacy contributed to the patient’s death. Though much less likely, and given the lack of a specific etiology for the patient’s death, the M.O. cannot completely rule out a toxic effect due to cefepime.

115

Patient AI411-204-005-472 was a 41 yo White female from the USA with multiple myeloma with CNS involvement. She received reinduction chemotherapy with cytarabine and daunorubicin from 1/13-1/19/94 followed by maintenance chemotherapy with methotrexate on 1/18/94 and radiation therapy. On 1/20/94, she developed febrile neutropenia (Temp=38.4 C, ANC=100, PLAT=82K), was enrolled, and received cefepime 2g IV q8h for 10 days (1/20 1/29/94). Baseline blood and urine cultures were negative. Baseline physical exam was notable for tachycardia, generalized edema, and oral mucositis. Pertinent Med Hx included: subdural hematoma in 1993, vertebral compression fractures, and depression. Pertinent concomitant therapy included: amphotericin B, ceftazidime (1/29-2/1), vancomycin (1/29-2/3), clindamycin (1/30-2/1), imipenem (2/1-2/4), amikacin (2/1-2/4), nystatin powder (1/31-2/1), silver sulfadiazine (1/31-2/4), ranitidine, amitriptyline, morphine, acetaminophen, diphenhydramine, allopurinol, Lasix, prochlorperazine, intrathectal methotrexate, anicaproic acid, dopamine (2/1 2/4), platelets and PRBCs. The hospital course was notable for the following: amphotericin was added on 1/25/94 (study day 6) for C. albicans isolated from urine cultures on Days 2 and 6 and a stool culture from Day 5. Methicillin-resistant coagulase-negative Staphylococci was isolated from blood cultures on Days 6-8 (1/25-1/27/94), and alpha-hemolytic Streptococci and Diphtheroids were isolated from a skin lesion on 1/28/94 (Day 9). The patient was withdrawn from the study on 1/29/94 (Day 10) due to persistent fever, tachycardia, and falling blood pressure and started on ceftazidime and vancomycin due to “breakthrough” bacteremia thought to be due to methicillin-resistant coagulase-negative Staphylococci. Other post-treatment antibacterials included imipenem and amikacin. On 2/3/94, the patient’s condition deteriorated and she developed hypothermia (Temp=35.4 C) and hypotension (SBP=60). She died on (b) (6) (b) (post therapy day (6) . No autopsy was performed. Serious adverse events reported were hypotension and sepsis deemed unrelated to cefepime. The investigator attributed the death to her multiple myeloma, neutropenia, acute plasma cell leukemia, and coagulase-negative Staphylococcal bacteremia and not to cefepime.

M.O. comment: This patient’s death may have been due to multiple factors including acute leukemia, multiple myeloma with central nervous system involvement, persistent neutropenia, and/or sepsis due to either an alpha-hemolytic Streptococcus or methicillin-resistant coagulase-negative Staphylococci. Methicillin-resistant coagulase-negative Staphylococci and many alpha-hemolytic Streptococci may not be susceptible to cefepime; therefore, this patient’s death may not have been due to a lack of cefepime efficacy. The M.O. cannot completely rule out the possibility that a toxic effect due to cefepime may have contributed to the patient’s death.

Patient AI411-204-005-475 was a 48 yo Asian female from the USA with chronic myelogenous leukemia in blast crisis. She received reinduction chemotherapy with cytarabine and daunorubicin from 1/16-1/22/94. On 1/25 she developed febrile neutropenia (Temp=38.7 C, ANC=36, PLAT=17K) with cough and lethargy, was enrolled, and received cefepime 2gIVq8h for 9 days (1/26-2/3/94). Baseline physical exam revealed abdominal cramping and nausea. Baseline blood cultures x2 yielded K. pneumoniae. Pertinent Med Hx included: rectal adenocarcinoma in 1986 s/p resection, adjunctive chemotherapy, and colostomy placement; and hematuria due to hemorrhagic cysts. Pertinent concomitant therapy included: Flagyl (1/31-2/3 and 2/6-2/7), amphotericin B (1/31-2/8, 2/10-2/11), ceftazidime (2/3-2/7), clindamycin (2/3-2/6), oral vancomycin (2/3-2/12), clotrimazole troche (2/1-2/9), vancomycin IV (2/6-2/9), gentamicin

116

(2/6 only), imipenem (2/7-2/12), 5-flucytosine (2/9 only), tobramycin (2/11 only), Percocet, ondansetron, Tylenol, Benadryl, lasix, meperidine, ranitidine, dopamine, anicaproic acid, atropine, Ativan, KCl, codeine, PRBCs, platelets, FFP. During the hospital course, repeat blood cultures on 1/26, 1/31, 2/1, and 2/2 were negative for growth, then on 2/4 and 2/5 blood cultures were positive for E. faecium. Also on 2/2 bronchial washing culture was positive for Penicillium species. Assays for C. difficile toxin were positive on 1/26 and 1/31. Serious adverse events (b) reported included: GI bleeding starting on 2/2/94 and lasting (6) days (to death) and respiratory (b) failure starting on 1/31 and lasting (6) days (to death), jaundice (1/31-death), pleuritic chest pain (2/2-death), blood blisters in mouth (2/4-death). None of the SAEs were attributed to cefepime. At an unclear point in the patient’s hospital course, her condition deteriorated and was transferred to the ICU due to hypotension and respiratory difficulty. The patient died on (b) (6) The investigator deemed that the patient’s death was due to cardiac dysrhythmia, hyperkalemia, acute renal failure, CML in blast crisis, and E. faecium bacteremia. (Baseline potassium and creatinine were 3.1 and 0.5, respectively. BMS did not provide results for these lab values beyond 2/3/94.)

M.O. comment: This patient’s death was likely due to multiple factors including E. faecium sepsis, cardiac arrhythmia, hyperkalemia, acute renal failure, and CML in blast crisis. Cefepime is not effective against E. faecium. The M.O. cannot rule out that cefepime may, in part, have played a role in the patient’s renal failure; however, additional factors may have also contributed to the renal failure: septic shock, blast crisis, and antimicrobials (aminoglycoside, amphotericin B, vancomycin, etc.).

Patient AI411-204-005-559 was a 62 yo Black male from the USA with acute lymphocytic leukemia (ALL). He received reinduction chemotherapy with cytarabine and mitoxantrone from 7/27-8/2/94. He developed febrile neutropenia (Temp=39.9 C, ANC=100) and received cefepime 2gIVq8h for 11 days (8/4-8/14/94). Baseline blood and urine cultures were negative. Pertinent Med Hx included: congestive heart failure, COPD, and esophageal stricture. Pertinent concomitant therapy included: metronidazole (8/8-8/11), vancomycin (8/14-8/30), amphotericin B (8/27-8/31), imipenem (8/14-8/25), fluconazole (8/15-8/26), nifedipine, insulin, acetaminophen, prochlorperazine, diphenoxylate, potassium and magnesium replacement, PRBCs, and platelets. Metronidazole was added on Day 5 (8/8) for Trichomonas infection. E. faecium was isolated from blood cultures on 8/11 and 8/14 resulting in discontinuation of cefepime. The patient deteriorated, was transferred to the ICU, and was placed on imipenem, vancomycin, fluconazole, and amphotericin. T. glabrata was isolated from blood on 8/22 and Enterobacter sp. from sputum on 8/28. The patient died on (b) (6) days post therapy). The investigator deemed that the patient’s death was caused by progression of acute leukemia, metabolic acidosis, T. glabrata fungemia, and pneumonia and unrelated to cefepime. The following were notable AEs: prolonged diarrhea (8/7-death), metabolic acidosis (8/13-death), and altered mental status and pulmonary congestion (8/14-death). Baseline creatinine was 1.2 and rose to 2.1 on 8.14 (last lab value noted in CRF). None of these AEs were considered related to cefepime per investigator.

M.O. comment: This patient’s death was likely due to multiple factors including fungemia and sepsis, progression of acute leukemia, progressive renal dysfunction, and metabolic

117

acidosis. The M.O. cannot rule out that cefepime may have contributed to the renal dysfunction.

Patient AI411-204-007-022 was a 23 yo White female from the USA with adenocarcinoma of the lacrimal gland with metastases to the brain and bones. She received maintenance chemotherapy with MESNA, doxorubicin, ifosfamide, with dacarbazine that ended on 2/15/93. On 2/23, she developed febrile neutropenia (Temp=39.1 C, ANC=0, PLAT=29K) with tachycardia, lethargy, stomatitis, and nausea and was placed on cefepime 2gIVq8h for 9 days (2/24-3/4/93). Baseline blood and urine cultures were negative. Pertinent Med Hx included: mental status changes and seizures related to brain metastases. Pertinent concomitant therapy included: fluconazole (2/25-3/4), norpramine, carafate slurry, Dilantin, vitamin K, Decadron, G CSF, compazine, Ativan, Tylenol, valium, dilaudid, benadryl, potassium and magnesium replacement, PRBCs, and platelets. Hospital course was notable for worsening mental status reportedly due to worsening brain metastases. The patient defervesced on 2/25/93 and was discharged on (b) (6) . She was re-admitted on (b) (6) due to grand mal seizures related to brain (b) (b) (6) metastases. The patient died(6) days post therapy on . The investigator deemed that the patient’s death was caused by worsening metastatic adenocarcinoma and unrelated to cefepime. The following were notable AEs: increased PTT coagulopathy (2/28-3/8) of moderate severity and deemed related to cefepime therapy which resolved after stopping cefepime and giving vitamin K; facial droop and change in mental status (2/25 only) of moderate severity and deemed unrelated to cefepime; inability to follow commands/change in neurologic status (3/1-3/23) of mild severity that appeared to improved 19 days after cefepime was stopped deemed as having an unknown relationship to study drug; severe thrombocytopenia (2/24-3/4) which abated after cefepime was stopped and deemed to have an unknown relationship with cefepime; and mild hypokalemia and moderate anemia deemed to have unknown relationships with cefepime therapy.

M.O. comment: This patient’s death was likely due adenocarcinoma with metastases to the brain and bones. Seizures were likely due to brain metastases, however; the M.O. cannot completely rule out that cefepime may have contributed to the seizures. The M.O. cannot rule out that cefepime may caused or contributed to the increased PTT and thrombocytopenia. The patient’s underlying disease may have also been associated in increased PTT and thrombocytopenia.

Patient AI411-204-010-070 was a 45 yo White female from the USA with non-Hodgkin’s lymphoma. She received reinduction chemotherapy with VP-16, cisplatin, and Adriamycin from 9/25-9/28/93. She developed renal failure during chemotherapy and a cystogram revealed ureteral obstruction due to a bladder mass requiring stent placement on 9/25. On 10/6/93 she experienced febrile neutropenia (Temp=38.2 C, ANC=376, PLAT=61K) with tachycardia, weakness, and lower extremity edema. She was placed on cefepime 2gIVq8h for 4 days (10/6 10/9). Baseline blood and urine cultures were negative. Pertinent Med Hx included: obesity, pleural effusions, dyspnea, nausea, vomiting, ascites, hyperkalemia, hypokalemia, hypocalcemia, hypomagnesemia, and codeine allergy. Pertinent concomitant therapy included: Ancef (10/9 only); Reglan; G-CSF; allopurinol; zantac; phenergan; potassium, magnesium, and calcium supplementation; Zofran; Decadron; morphine; Marinol; Tylenol; Propofol; fentanyl, droperidol, and lasix. Cefepime was discontinued on 10/9 because her fever persisted and the investigator

118

reportedly felt that it (the fever) was not due to a bacteriologic cause. Even so, the patient was placed on Ancef for 1 day and the patient remained febrile. She was discharged on (b) (6) and was readmitted on (b) (6) with dyspnea, tachycardia (HR=158), and fever (38.3 C). Blood and urine cultures were positive for E. faecium. She was treated with ampicillin and ciprofloxacin and the infection reportedly cleared by 10/17, however; low grade fever persisted and was believed to be due to worsening lymphoma that was confirmed by CT scan. She was placed on comfort care measures and died on (b) (6) days post therapy). The investigator deemed that the patient’s death was caused by Non-Hodgkin’s lymphoma. The following were notable AEs during the course of the study: thrombus in right subclavian vein/portacath (10/8) deemed of mild severity and unrelated to cefepime and resolved when the portacath was removed; sepsis of moderate severity (10/15-10/17) deemed unrelated to study drug and prolonged the hospital stay; mild edema (10/15-10/17) deemed unrelated to study drug and resolved with lasix, mild muscle twitching (10/7 only) deemed unrelated to study drug and resolved with calcium supplementation.

M.O. comment: This patient’s death was likely due to advanced lymphoma.

Patient AI411-204-010-133 was a 65 yo White male from the USA with lung cancer (affecting the right middle and upper lobes) and metastatic liver and bone lesions. He received radiation therapy from 10/22-11/4/93 and chemotherapy with Velban on 11/9/93. On 11/17/93, he developed febrile neutropenia (Temp=39.3 C, ANC=295) with cough, dyspnea, tachycardia, chills, and malaise, and was placed on cefepime 2gIVq8h for 5 days (11/17-11/21/93). Baseline blood cultures x4, sputum, and urine cultures were negative. CXR demonstrated a right lower lobe lesion. However, initially, the treating physician felt that the patient’s infection source was a skin ulcer on the buttock that grew methicillin-susceptible S. aureus and E. coli. Pertinent Med Hx included: pancytopenia, allergy to Neosporin, and tobacco use. Pertinent concomitant therapy included: vancomycin 2 g/day (11/19-11/26), mezlocillin 12 g/day (11/21 to death), ceftazidime 3 g/day (11/21 to death), erythromycin (11/21 to death), Diflucan (11/24 to death), Tylenol, compazine, G-CSF, Proventil, Vanceril, Valium, Thorazine, hydrocortisone, Pepcid, lasix, benadryl, potassium supplementation, Versed, Ativan, Haldol, Dilaudid, Reglan, PRBCs, and platelets. Over the next 2 days, the patient’s fever worsened and CXR demonstrated bilateral pulmonary infiltrates and vancomycin was added. On 11/21 the patient was transferred to the ICU and was placed on mechanical ventilation for respiratory failure. The cefepime was discontinued and replaced with ceftazidime, mezlocillin, and erythromycin. Fluconazole was added on 11/22 because of Candida sp. in the sputum. According to page 53 of the CRF, “CXR had improved considerably by 11/23/93 but pt continued with increased FIO2 requirement (PO2 61 on FIO2 .60 & PEEP 10) and decreased SaO2 with any stress/stimulation. Oncologist at this point felt that cancer was end-stage; family requested discontinuation of life support/therapy on 11/26/93.” The patient died on (b) (6) days post cefepime). No autopsy was performed. The investigator deemed that the patient’s death was caused by the underlying lung cancer and Candida pneumonia, and was not due to cefepime. The following SAEs were noted during the course of the study: hypoxia (11/21/93 to death) and respiratory failure (11/22 to death) were deemed unrelated to study therapy.

M.O. comment: This patient’s death appeared to ultimately be due to advanced lung cancer.

119

The differential diagnosis for the patient’s lung lesions includes: lung cancer, inflammatory response to chemotherapy, pulmonary edema, Candidal pneumonia, and bacterial pneumonia.

Patient AI411-204-010-186 was a 51 yo White female from the USA with breast cancer and liver metastases. She received maintenance chemotherapy with carboplatin and VP-16 on 5/9/94. On 5/17, she developed febrile neutropenia (Temp=100.6 F, ANC=18) with cough and nasal drainage, and was placed on cefepime 2gIVq8h for 4 days (5/17-5/20/94). Baseline cultures of blood x2, urine, and pleural effusion were negative. Pertinent Med Hx included: CVA with left hemiparesis and disorientation on 5/9/94, carotid artery stenosis, cardiomyopathy, pancytopenia, tobacco use, and pleural effusions. Pertinent concomitant therapy included: lanoxin, lasix, capoten, megace, aspirin, prilosec, G-CSF, dilaudid, cisapride, Reglan, dilaudid, benadryl, Tylenol, lasix, phenergan, Inapsine, Ativan, potassium supplementation, PRBCs, platelets. Pertinent laboratory results included: serum creatinine at baseline and on 5/20 were 0.5 and 0.6, respectively. The patient defervesced and cefepime was discontinued on 5/20. She was discharged to hospice care on (b) (6) The breast cancer progressed and the patient died on (b) (6) (b) ( (6) days post cefepime). No autopsy was performed. The investigator deemed that the patient’s death was caused by underlying metastatic breast cancer and was unrelated to cefepime. The following was a notable AE during the course of the study: mild expressive aphasia (5/19-5/21) deemed to be related to concomitantly administered narcotics and unrelated to cefepime.

M.O. comment: This patient’s death was likely due to advanced metastatic breast cancer. The M.O. cannot rule out the possibility that the patient’s transient, mild expressive aphasia may have been related, at least in part, to cefepime exposure given that the event occurred while on cefepime.

Patient AI411-204-010-366 was a 46 yo White female from the USA with disseminated Non Hodgkin’s lymphoma (including meningeal, liver, adrenal, and bilateral kidney involvement). She received chemotherapy from 8/21-8/30 with cisplatin, Ara-C, and methotrexate. On 9/1/93, she developed febrile neutropenia (Temp=40.4 C, ANC=5) with hypotension (SBP=93), altered mental status, and urinary incontinence. Baseline cultures yielded K. pneumoniae in the blood and E. aerogenes in the urine. She was placed on cefepime 2g x 1 dose (9/1/93). Pertinent Med Hx included: oral candidiasis, HTN, hepatomegaly, depression, panic attacks, and penicillin allergy. Pertinent concomitant therapy included: Flagyl (8/29-8/31), Diflucan (9/1 to death), Pepcid, Decadron, Zofran, dilaudid, restoril, Lomotil, PPN nutrition, Intralipid, magnesium supplementation, PRBCs, platelets, G-CSF, and additional antibacterial agents as listed below. Pertinent laboratory results included: serum creatinine was 1.9 at baseline and 3.1 on 9/4. On the same day that she was enrolled in the study (9/1), the patient was diagnosed with septic shock and was transferred to the ICU for pressor support and mechanical ventilation. Cefepime was changed to ceftazidime 4g/day, ciprofloxacin 800mg/day, and vancomycin 1g/day. Her vital signs reportedly stabilized over the next 48 hours however, her mental status worsened and she became less responsive. On 9/4, creatinine had worsened to 3.1. On 9/8, the family declined hemodialysis and elected to withdraw care due to disseminated lymphoma. The patient died on (b) (6) days post single dose of cefepime). No autopsy was performed. The investigator deemed that the patient died due to disseminated lymphoma, septic shock, and renal failure. Septic shock was considered an SAE unrelated to study therapy.

120

M.O. comment: This patient’s death was ultimately due to disseminated lymphoma complicated by septic shock and renal failure. The family elected to withdraw care due to the terminal nature of the lymphoma. The M.O. cannot rule out the possibility that cefepime failed to treat the pathogens causing septic shock; however, the patient only received 1 dose of cefepime and was switched to other antibacterial therapy within hours of enrollment.

Patient AI411-204-010-561 was a 72 yo White male from the USA with acute myelogenous leukemia. He received chemotherapy with Ara-C and Idarubicin from 7/19-7/26/94 and again from 8/4-8/11/94. On 7/27 he developed febrile neutropenia (Temp=38.7 C, ANC=8) and was placed on cefepime 2gIVq8h for 15 days (7/27-8/10). Baseline blood and urine cultures were negative. Pertinent Med Hx included: HTN, hypothyroidism, pancytopenia, and prostate cancer. Pertinent concomitant therapy included: Diflucan (7/19-8/6), oral ciprofloxacin 1 g/day (7/18 7/27), Sulamyol 10% topical ophthalmic solution (7/22-7/29), vancomycin (8/4 to death), allopurinol, hydrochlorothiazide, Synthroid, benadryl, Tylenol, lasix, dilaudid, Zofran, Reglan, potassium supplementation, Lomotil, SoluMedrol, Tylenol #3, PRBCs, and platelets. Pertinent laboratory results included: leukemic blast cells noted in blood differential, platelets=31 at baseline. On 7/29 (Study Day 3), the patient developed a maculopapular rash on his forearms and legs. On 8/10 (Study Day 15), the rash extended to the face and torso and cefepime was discontinued and was replaced with imipenem 2gm/day. The rash included exfoliation of the palms and soles and “resolved within a few days without treatment”. Of note, the patient had received a second course of Ara-C from 8/4-8/11. On 8/17 (post therapy day 7), the patient developed a new fever and blood cultures revealed C. krusei fungemia. Fluconazole was added and the Hickman catheter was removed. On 8/31, the patient developed confusion and vomiting and died on (b) (6) . No autopsy was performed. The investigator deemed that the patient’s death was caused by fungal sepsis and acute leukemia. The following were notable AEs during the course of the study: the SAE of refractory myelosuppression (8/22 to death) deemed unrelated to cefepime, the SAE of candidemia (8/17 to death) deemed unrelated to cefepime, the AEs of hives and skin rash to trunk and legs (7/28-8/17) deemed mild and unrelated to study therapy, the AE of exfoliative reaction on palms and soles (8/9-8/26) deemed of moderate severity and unrelated to cefepime, and the AE of blood in stool (8/7 only) deemed mild and unrelated to study therapy.

M.O. comment: This patient’s death was due to fungal sepsis and acute leukemia. The rash described would be consistent with a reaction to Ara-C; however, the M.O. cannot rule out the possibility that cefepime may have contributed to the patient’s rash.

Patient AI411-204-012-373 was a 57 yo White female from the USA who had acute myeloid leukemia. She received chemotherapy with idarubicin, Ara-C, and VP-16 for 4 days (9/24-9/27). She developed febrile neutropenia (Temp=37.9 C, ANC=76) on 9/30/93 and received cefepime 2gIVq8h for 10 days (10/1-10/10/93). Baseline blood and urine cultures were negative. Pertinent Med Hx included: diabetes, peptic ulcer disease, peripheral neuropathy, anemia, cardiovascular disease, on INH monotherapy for latent mycobacterial infection(?), and anxiety. Pertinent concomitant therapy included: vancomycin (10/5-10/16), fluconazole (unknown start date to10/12), acyclovir, isoniazid, ciprofloxacin 1.5 g/day (unknown start date to10/1), Flagyl (10/10 to death), amphotericin B (10/13 to death), imipenem (10/10 to death), Darvon, zantac, vitamin B6, coumadin, premarin, vistaril, G-CSF, Percocet, restoril, Demerol, benadryl, Tylenol,

121

IVIG, danazol, Lomotil, potassium and magnesium supplementation, Lasix, PRBCs, and platelets. Pertinent laboratory results included: baseline platelets=53. The patient continued to be febrile and repeat cultures were negative. Cefepime was discontinued on 10/10/93 because fevers were not improving. Bone marrow biopsy 7 days post therapy (10/17) demonstrated persistent acute leukemia. The patient died on (b) (6) days post therapy). No autopsy was performed. The investigator deemed that the patient’s death was caused by acute leukemia. The following was a notable AE during the course of the study: recurrent fever of moderate severity (10/9-10/17) was considered unrelated to study therapy.

M.O. comment: This patient’s death was most likely due to acute leukemia. Persistent fever may have been related to acute leukemia or reactivation of mycobacterial infection.

Patient AI411-204-014-117 was a 74 yo White male patient from the USA with esophageal carcinoma. He had received radiation therapy (1/24-2/11/94) and 5-FU (1/31-2/3/94). On 2/8/94, he developed febrile neutropenia (Temp=38.6 C, ANC=290) with weakness, dysphagia, and diarrhea and received cefepime 2gIVq8h for 4 days (2/8-2/11). Baseline blood and urine cultures were negative. Pertinent Med Hx included: COPD, HTN, pancytopenia, depression, and renal insufficiency. Pertinent concomitant therapy included: Tylenol, G-CSF, steroid cream, Lomotil, Lotensin, total parenteral nutrition, morphine, electrolyte supplementation, and PRBCs. Pertinent laboratory results included: creatinine: 1.9 on 2/8 and 2.6 on 2/15. The patient completed cefepime therapy on 2/11 with resolution of fever; however, on 2/13, during a second cycle of radiation therapy, he vomited 300 cc of black emesis and was found to have a GI bleed due to radiation-induced esophageal necrosis. Bloody emesis continued and on 2/14, the patient was transferred to the ICU due to hypotension and tachycardia and was found to have a new infiltrate on CXR thought to be due to aspiration pneumonia. On 2/15, C. freundii was isolated from sputum and he was placed on ceftizoxime, ciprofloxacin, and clindamycin. The patient continued to deteriorate and the decision was made to remove him from life support and provide comfort measures only. On (b) (6) days post cefepime), the patient died. No autopsy was performed. The investigator deemed that the patient’s death was caused by esophageal cancer, GI bleed, pulmonary and hemodynamic deterioration, and aspiration pneumonia and was unrelated to cefepime. The following were SAEs noted between 2/13 and death and none were considered related to study therapy: GI bleed, aspiration pneumonia, third-spacing of fluid, hypovolemia, metabolic acidosis, hypotension, tachycardia, tachypnea, worsening renal insufficiency, and agitation.

M.O. comment: This patient’s death was most likely due to underlying co-morbidity (esophageal cancer, GI bleed) and resultant aspiration pneumonia.

Patient AI411-204-016-102 was an 82 yo White female from the USA with small cell lung cancer. She received radiation therapy from 6/23-7/22/93 and chemotherapy with etoposide and cisplatin from 5/30 through enrollment. She was enrolled in the study on 8/19/93 with febrile (Temp=38.7C) neutropenia (ANC not performed; however WBC=0.4) with bacteremia and pneumonia and received cefepime 2gIVq12h for 12 days (8/19-8/30). Baseline blood cultures x 2 were positive for K. pneumoniae. Pertinent Med Hx included: HTN, UTI 3 weeks prior to enrollment, oral thrush, pancytopenia, polymyalgia rheumatica, generalized weakness, fall at home on 8/18/93, tobacco use, and glaucoma. Pertinent concomitant therapy included:

122

fluconazole, nystatin, Tylenol, Tylenol with codeine, cardizem, zantac, prednisone, vasotec, procardia, potassium and magnesium supplementation, procardia, albuterol, and PRBCs. Pertinent laboratory results included: serum creatinine was 1.8 on 8/19, improved to 1.2 on 8/23, then worsened to 2.3 on 8/30. During the course of the study, repeat blood cultures were negative on 8/21 and 8/23 and she was afebrile. On (b) (6) while attempting to ambulate to the bathroom unassisted, she fell and sustained a head injury to the right occipital region. After the fall, the patient’s mentation never returned to baseline and she progressed to a semi-comatose (b) state until death. Her family elected to transfer the patient to hospice care and she died (6) days post therapy on (b) (6) No autopsy was performed. The investigator deemed that the patient’s death was caused by underlying lung cancer. The following SAEs were noted during the course of the study: altered mentation post fall of severe severity (8/24 to death) was not deemed related to cefepime and the mentation did not improve off cefepime, increasing creatinine of moderate severity (8/30 to death) was deemed as having an unknown relationship with cefepime and was not assessed for improvement after the end of the study.

M.O. comment: This patient’s death was ultimately due to small cell lung cancer complicated by a fall and subsequent altered mentation that deteriorated to a semi-comatose state which led to the decision to withdraw all care except for comfort measures. The M.O. cannot rule out the possibility that cefepime may have contributed to the patient’s altered mentation and renal dysfunction.

Patient AI411-204-016-402 was a 43 yo White male from the USA with poorly differentiated nodular lymphoma. Recent chemotherapy included: fludarabine (4/93-7/93) and 2-CDA (8/93 through enrollment). On 9/13/93, he developed febrile neutropenia (Temp=39.2 C, ANC not performed; however WBC=0.3) and received cefepime 2gIVq8h for 4 days (9/13-9/16). One out of 4 baseline blood cultures was positive for E. faecium. Pertinent Med Hx included: macular rash on admission, herpes zoster infection 2 weeks prior to enrollment, dyspnea on exertion, pancytopenia, and tobacco use. Pertinent concomitant therapy included: antimicrobial agents as discussed below, acyclovir, G-CSF, hydrocortisone, benadryl, restoril, Tylenol, potassium and calcium supplementation, Ativan, PRBCs, and platelets. Pertinent laboratory results included: on 9/13 (creatinine=1.1, AST=733, ALT=846) and on 9/16-last day of recorded lab work- (creatinine=0.8, AST=270, ALT=486). On 9/16/93, the patient was discontinued from the study due to persistent fever and isolation of a resistant organism. His antibacterial therapy was changed to vancomycin, gentamicin, and cefoperazone. On 9/20, cefoperazone was discontinued due to suspected drug rash. Repeat CXR showed right middle and upper lobe infiltrates. Amphotericin was added empirically on 9/19. On 9/22, he was transferred to the ICU. On 9/24, open lung biopsy demonstrated Aspergillus pneumonia. Comfort care measures were initiated and the patient died on (b) (6) days post therapy). No autopsy was performed. The investigator deemed that the patient’s death was caused by persistent pancytopenia secondary to lymphoma and Aspergillus pneumonia and was not related to cefepime. The following was a notable SAE during the course of the study: sepsis exacerbation (9/16 to death) was deemed unrelated to cefepime.

M.O. comment: This patient’s death was most likely due to Aspergillus pneumonia, lymphoma, and pancytopenia.

123

Patient AI411-204-016-455 was a 37 yo White male from the USA with chronic myelogenous leukemia in blast crisis. Recent chemotherapy included: Hydrea (2/91 to 12/13/93). He developed febrile neutropenia (Temp=39.3 C, ANC=0) on 12/13/93 and received cefepime 2gIVq8h for 9 days (12/13-12/21). Physical exam revealed tachycardia, tachypnea, and epistaxis. Baseline blood and urine cultures were negative. Pertinent Med Hx included: pancytopenia, prolactinoma, psoriasis, epistaxis, visual field loss in the left eye, and elevated liver function tests. Pertinent concomitant therapy included: bacitracin, Parlodel, Tylenol, Demerol, Lasix, electrolyte supplementation, Zofran, and 10 units of PRBCs and 72 units of platelets between 12/13 and 12/21/93. Pertinent laboratory results included: at baseline (12/13): hemoglobin=6.5, platelets=25, creatinine=1.7; on 12/21 (last recorded): hemoglobin=9.1, platelets=17, creatinine=0.7. On Study Day #1, the patient started reinduction chemotherapy (Ara-C and daunorubicin) and was prophylactically placed on fluconazole (100 mg/day) and norfloxacin (800 mg/day). The hospital course was complicated by epistaxis requiring nasal packing and gel foam. By (b) (6) , the patient was afebrile, cefepime was discontinued, and the patient was discharged home. On (b) (6) days post therapy), the patient was readmitted with fever, pancytopenia (WBC=500, platelets=1,000), and coagulopathy. The patient was empirically placed on tobramycin and cefoperazone. The patient was hypotensive (SBP=88) and required dopamine. Physical exam demonstrated multiple sites of bleeding: gingival, retinal, epistaxis. Head CT, obtained due to headache, dizziness, and nausea, showed cerebral hemorrhage. Hemoglobin did not improve despite multiple transfusions. On(b) (6) the patient (b) became profoundly hypotensive, developed asystole, and died (6) days post cefepime). No autopsy was performed. The investigator deemed that the patient’s death was caused by cerebral hemorrhage secondary to thrombocytopenia secondary to leukemia and unrelated to cefepime. The following were notable AEs during the course of the study: epistaxis (12/18-12/19) of moderate severity that was deemed unrelated to cefepime; cerebral hemorrhage (SAE) (1/4/94 to death) deemed unrelated to cefepime.

M.O. comment: This patient’s death was most likely due to cerebral hemorrhage secondary to prolonged thrombocytopenia secondary to leukemia in blast crisis and concomitant chemotherapy. The M.O. cannot completely rule out the possibility that cefepime may have contributed to the prolonged thrombocytopenia; however, the other factors mentioned above likely played a major role in causing the prolonged thrombocytopenia.

Patient AI411-204-016-490 was an 18 yo White female from the USA with acute myelogenous leukemia. Recent chemotherapy (4/19-4/27/94) included: VP-16, daunomycin, Ara-C, and cyclosporine. She developed febrile neutropenia (Temp=38.9 C, ANC not performed; however WBC=0.2) on 4/29/94 and received cefepime 2gIVq8h for 9 days (4/29-5/7/94). Baseline blood and urine cultures were negative. Pertinent Med Hx included: increased LFTs, hepatomegaly, loose stools, pancytopenia, bone marrow transplant 12/2/93 followed by relapse of leukemia on 3/25/94, and peripheral neuropathy. Pertinent concomitant therapy included: fluconazole (4/18 5/9/94), norfloxacin (4/18-4/28), vancomycin (5/2-5/12), imipenem (5/7-5/12), amphotericin B (5/9-5/12), acyclovir, Tylenol, Restoril, benadryl, morphine, hydrocortisone, Darvocet, Ativan, desmopressin, decadron, insulin, thiamine, mannitol, esmolol, propofol, Dilantin, midazolam, dopamine, 18 units of platelets and 4 units of PRBCs. Pertinent laboratory results included: at baseline (4/28): hemoglobin=9.0, platelets=28, creatinine=0.5; on 5/8 (last recorded): hemoglobin=10.2, platelets=13, creatinine=0.4. The patient remained febrile despite repeatedly

124

negative blood cultures. On 5/3, C. difficle infection was diagnosed, treated with oral vancomycin, and was deemed to probably be related to cefepime therapy. On 5/1 the patient began developing headaches that progressed to altered mentation by 5/8. Head CT demonstrated a left basal ganglion mass. The patient continued to deteriorate developing decorticate posturing and died on (b) (6) days post cefepime). No autopsy was performed. The investigator deemed that the patient’s death was caused by left basal ganglion mass (etiology of mass unknown because family declined autopsy) and recurrent leukemia and unrelated to cefepime. The following were SAEs noted during the course of the study: headache (5/1 to death), blurred vision (5/8-5/9), neurological changes due to cerebral mass (5/9 to death), and hypotension (5/11 to death). These SAEs were deemed unrelated to cefepime.

M.O. comment: This patient’s death was most likely due to cerebral mass and acute leukemia. The differential diagnosis for the etiology of the cerebral mass may include leukemia versus infection/abscess. If the mass was due to a susceptible bacterial infection, then cefepime would potentially be implicated for treatment failure; however, the family declined autopsy and all blood cultures were reported as negative. Cefepime may have caused the C. difficle infection.

Patient AI411-204-016-541 was a 45 yo White male from the USA with acute myelogenous leukemia. Recent chemotherapy (5/2-5/8) included: idarubicin and Ara-C. On 5/5/94, he developed febrile neutropenia (Temp=39.7 C, ANC=290) and received cefepime 2gIVq8h for 4 days (5/5-5/8). Physical exam revealed dyspnea with bilateral rales, non-productive cough, peri rectal erythema, and fundi with bilateral hemorrhage. Baseline blood and urine cultures were negative. Pertinent Med Hx included: MI in 1986, GERD, pancytopenia, hyperglycemia, Agent Orange exposure, and a 12-pack of beer three times/week. Pertinent concomitant therapy included: imipenem (5/8 to death), tobramycin (5/8-5/12), fluconazole, norfloxacin (5/3 to 5/5), allopurinol, Tylenol, lasix, nitroglycerin, codeine, zantac, captopril, morphine, esmolol, diltiazem, digoxin, versed, procainamide, insulin, methylprednisolone, amiodarone, dopamine, electrolyte supplementation, PRBCs and platelets. On 5/6, the patient was transferred to the ICU due to hypoxia and hypotension (SBP=78). On 5/8, cefepime was changed to imipenem and tobramycin. Open lung biopsy demonstrated areas of acute lung infarction, chemotherapeutic drug toxicity was deemed the most likely etiology and Solumedrol was started. On 5/17, vancomycin and amphotericin B were added. Renal failure was noted on 5/19 with creatinine=2.7. On (b) (6) days post therapy) the patient died of respiratory failure. No autopsy was performed. The investigator deemed that the patient’s death was caused by respiratory failure secondary to chemotherapy toxicity, and persistent neutropenia. The following were notable AEs during the course of the study: respiratory distress (SAE) (5/6 to death), atrial tachycardia (SAE) (5/9 to death), and hyperglycemia (SAE) (5/7 to death). All of these SAEs were deemed unrelated to cefepime.

M.O. comment: This patient’s death was most likely due to respiratory failure due to acute lung injury from chemotherapy for leukemia. The M.O. cannot rule out that cefepime may, in part, have played a role in the patient’s renal failure; however, additional factors may have also contributed to the renal failure: leukemia, hypotensive episodes, chemotherapy, and antimicrobials (aminoglycoside, amphotericin B, vancomycin, etc.).

125

Patient AI411-204-018-233 was a 48 yo White female from the USA with metastatic breast cancer (metastases to lungs, liver, bone marrow, CNS). Recent chemotherapy included: radiotherapy 8/15-9/5/94 to the brain and chemotherapy (9/12-9/14) with mitoxantrone, leucovorin, 5-FU, and intrathectal thiotepa. On 9/21/94, she developed febrile neutropenia (Temp=38.0 C, ANC=200) and received cefepime 2gIVq8h for 12 days (9/22-10/3/94). Baseline blood and urine cultures were negative. Pertinent Med Hx included: pancytopenia, allergy to penicillin (mild rash), hypokalemia, hypophosphatemia, thrush, bilateral metastatic pleural effusions, and right pleurocentesis (9/13/94). Pertinent concomitant therapy included: nystatin, G-CSF, megace, electrolyte supplementation, Tylenol, multivitamin, Anusol, PRBCs, and platelets. Pertinent laboratory results included: serum creatinine was 0.5 at baseline (9/22) and 0.5 on 10/3 (last value recorded in CRF). On 10/3, cefepime was discontinued because the patient was afebrile and no longer neutropenic. The patient was discharged on (b) (6) At home, the patient’s respiratory status continued to deteriorate with dyspnea and increase pleural effusions, and dehydration. She was re-admitted for supportive care and expired on (b) (6) days post therapy). No autopsy was performed. The investigator deemed that the patient’s death was caused by pulmonary failure secondary to metastatic breast cancer and unrelated to cefepime. The following was a notable AE during the course of the study: depressed pulmonary function (10/1 to death) of moderate severity was deemed unrelated to cefepime.

M.O. comment: This patient’s death was most likely due to respiratory failure secondary to metastatic breast cancer.

Study AI411-204 (Comparator = ceftazidime):

Patient AI411-204-005-395 was a 61 yo White female from the USA with large cell lymphoma. Recent chemotherapy (9/15/93) included: cyclophosphamide and cisplatin. On 9/24/93, she developed febrile neutropenia (Temp=39.0 C, ANC=100) with cough, lethargy, and tachycardia and received ceftazidime 2gIVq8h for 9 days (9/24-10/2/93). Baseline blood cultures were negative. Pertinent Med Hx included: mild tricuspid and mitral regurgitation, pancytopenia, stomatitis, and anorexia. Pertinent concomitant therapy included: clotrimazole, fluconazole, allopurinol, acetaminophen, morphine, Benadryl, Kao-Pectin, electrolyte supplementation, Lasix, compazine, Roxicet, PRBCs, platelets, enteral nutrition. Pertinent laboratory results: creatinine=0.5 on 9/22/93 and 0.4 on 10/8. On 9/28/93 (study day 5), the patient defervesced; however, the patient’s clinical condition progressively deteriorated. She was discharged on (b) (6) and expired on (b) (6) days post therapy). No autopsy was performed. The investigator deemed that the patient’s death was caused by large cell lymphoma and was not related to study therapy. The following were notable AEs: mental status changes (9/24-10/3) of moderate severity were deemed unrelated to ceftazidime and resolved without remedial therapy; deterioration from large cell lymphoma (SAE) ((b) (6) only) resulted in death on same day and was deemed unrelated to study therapy.

M.O. comment: This patient’s death was most likely due to large cell lymphoma.

Patient AI411-204-005-471 [a.k.a. AI411-204-005-393] was a 38 yo White female with acute myelogenous leukemia followed by acute lymphoblastic leukemia. [The patient had successfully completed a 12 day course of cefepime 2gIVq8h (9/2-9/13/93) for a prior episode of febrile

126

neutropenia in the same study.] She received re-induction chemotherapy with mitoxantrone, vincristine, L-asparaginase, prednisone, and cyclophosphamide from 1/4-1/21/94. On 1/14/94, the patient developed febrile neutropenia (Temp=40.2 C, ANC=12) and received 5 days of ceftazidime 2g (1/14-1/18/94). Baseline blood cultures were positive for E. cloacae and K. pneumoniae. Pertinent Med Hx included: anthracycline-induced cardiomyopathy, hepatosplenic candidiasis, prior AML, anxiety, depression, and retinal hemorrhage. Pertinent concomitant therapy included: amphotericin B, flucytosine, Buproprion, sucralfate, acetaminophen, Benadryl, colace, Lorazepam, Solumedrol, dobutamine, Lasix, PRBCs, platelets, and electrolyte supplementation. On the same day, the patient’s condition deteriorated to septic shock and the following additional antimicrobials were added: tobramycin, vancomycin, aztreonam, and later cefotaxime. During the hospital course, the patient reportedly developed a concomitant “staphylococcal bacteremia” (species and microbiologic susceptibilities not provided in CRF). The patient died on (b) (6) The investigator deemed that the patient’s death was caused by sepsis syndrome and ALL. No autopsy was performed. The following SAE was noted in the CRF: hypotension (1/14/94 to death)

M.O. comment: This patient’s death was most likely due to sepsis syndrome related to the baseline Gram-negative pathogens and/or secondary staphylococcal bacteremia and underlying acute lymphoblastic leukemia. The M.O. cannot rule out study drug failure with ceftazidime.

This patient was originally listed as a survivor for the cefepime study arm (Patient AI411-204- 005-393) according to BMS; however, the patient actually died while on ceftazidime during a subsequent episode of febrile neutropenia, therefore, the M.O. has re-assigned the patient to the ceftazidime arm of Study 204 for the mortality analysis.

Patient AI411-204-005-491 was a 54 yo White male from the USA with acute myelogenous leukemia. Recent chemotherapy (2/28-3/9/94) included: hydroxyurea and cytarabine. On 3/9/94, he developed febrile neutropenia (Temp=39.2 C, ANC<100) and received ceftazidime 2gIVq8h for 6 days (3/9-3/14/94). Physical exam revealed tachycardia, tachypnea, penile lesion, thrush, mucositis, bloody diarrhea, and abdominal pain. CXR was negative for infectious source. Baseline cultures as follows: blood cultures x2 were positive for alpha- hemolytic Streptococci, penile lesion culture was positive for C. freundii and P. mirabilis. Pertinent Med Hx included: cerebellar toxicity due to cytarabine, pancytopenia, diarrhea, and Aspergillus otitis externa. Pertinent concomitant therapy included: octreotide, morphine, electrolyte supplementation, lorazepam, acetaminophen, benadryl, insulin, ondansetron, dexamethasone, allopurinol, meperidine, promethazine, platelets and PRBCs. Pertinent laboratory values included: creatinine= 1.0 on 3/9 and 3/21; total bilirubin=1.2 on 3/9 and 14.8 on 3/21. Vancomycin IV and PO were added on 3/10 due to persistent fever and the potential for C. difficile infection. Itraconazole was added next. Imipenem was added on 3/14 and continued to 3/25. Repeat blood cultures were negative until on 3/19 and 3/20 when C. guillermondii was isolated from blood cultures. Amphotericin B was started on 3/19. On 3/25, the patient was (b) (6) (b) transferred to the ICU due to respiratory failure and died on (post therapy day (6) . No autopsy was performed. The investigator deemed that the patient’s death was caused by respiratory failure, C. guillermondii fungemia, neutropenia, and leukemia and was unrelated to study drug. The following SAEs were noted in the CRF: extremity edema (3/10 to death)

127

deemed unrelated to study therapy; and jaundice (3/14 to death) deemed unrelated to study therapy.

M.O. comment: This patient’s death was most likely due to C. guillermondii fungal sepsis due to prolonged neutropenia secondary to acute leukemia.

Patient AI411-204-010-067 was a 78 yo White male from the USA with disseminated GI adenocarcinoma (metastases to lung and liver). He received radiotherapy 6/3-6/15/93 and chemotherapy (7/31-8/1/93) with 5-FU and VP-16. On 8/9/93, he developed febrile neutropenia (Temp=39.9 C, ANC=110) and received ceftazidime 2gIVq8h for 7 days (8/9-8/15/93). Baseline culture results follow: blood grew methicillin-resistant coagulase-negative Staphylococci; urine grew S. marcescens; and a buttock ulcer exudate grew P. aeruginosa and B. fragilis. Pertinent Med Hx included: arrhythmia requiring cardioversion, and DVT in 1988. Pertinent concomitant therapy included: Lanoxin, heparin, carafate, prilosec, G-CSF, Lasix, compazine, Ativan, Tylenol, electrolyte supplementation, and morphine. Vancomycin was added on 8/11 (Day 3) to cover methicillin-resistant coagulase-negative Staphylococci, and the patient defervesced on the same day. The patient was discharged home on (b) (6) for comfort care and expired on (b) (6) (b) (6) days post therapy). No autopsy was performed. The investigator deemed that the patient’s death was caused by disseminated adenocarcinoma and was unrelated to study therapy. The following SAE was noted in the CRF: disseminated adenocarcinoma (8/25 to death) was deemed unrelated to study therapy.

M.O. comment: This patient’s death was most likely due to widely metastatic adenocarcinoma.

Patient AI411-204-010-135 [a.k.a. AI411-204-010-132] was a 72 yo White male with esophageal adenocarcinoma with hepatic metastases. [The patient had successfully completed a 6 day course of cefepime 2gIVq8h (11/5-11/9/93) for a prior episode of febrile neutropenia in the same study.] He received maintenance chemotherapy with VP-16, Adriamycin, and cisplatin from 11/16-11/18/93. On 11/25/93, the patient developed febrile neutropenia (Temp=39.6 C, ANC=144) with chills, malaise, tachycardia, and dyspnea and the patient was placed on ceftazidime 2g IV q8hr for 3 days (11/25-11/27 [death]). Baseline blood cultures x2 were positive for C. septicum. C. septicum was also isolated from a right elbow cellulitis aspirate. Pertinent Med Hx included: CAD with MI, emphysema, pancytopenia, anemia, hypothyroidism, and smoker (½ to 1 ppd). Pertinent concomitant therapy included: phenergan, Tylenol, FeSO4, Isordil, Synthroid, Pepcid, Quinine, Immodium, Tylenol with codeine, Darvocet, Maalox, and Mylicon. The patient’s condition deteriorated to septic shock and on (b) (6) mezlocillin and vancomycin were added. The patient died later that same day on (b) (6) . The investigator deemed that the patient’s death was caused by advanced adenocarcinoma, myelosuppression, and C. septicum sepsis. No autopsy was performed. AEs of mild indigestion and intestinal bloating and moderate emesis were noted during therapy with cefepime, however; no AEs were noted in the CRF for the period of ceftazidime therapy.

M.O. comment: This patient died due to C. septicum septic shock. The M.O. cannot rule out study drug failure with ceftazidime. According to the “CRF Discrepancy/Resolution Form” at the end of the CRF, this patient #132 was the same as patient #135. In the protocol, a patient

128 was apparently allowed to re-enroll into Study 204 seven days after completion of prior study therapy. The patient re-entered the study on ceftazidime 18 days post cefepime therapy.

This patient was originally listed as a cefepime death according to BMS; however, the patient actually died while on ceftazidime, therefore, the M.O. has re-allocated the patient to the ceftazidime arm of Study 204.

Patient AI411-204-010-188 was a 50 yo White female from the USA with metastatic breast cancer (metastases to the liver, bone, and lung with bilateral effusions). Recent radiotherapy occurred from 12/13-12/28/93. Recent chemotherapy (6/3-6/5/94) included: Taxol. On 6/7/94, she developed febrile neutropenia (Temp=38.6 C, ANC=399) and received ceftazidime 2gIVq8h for 12 days (6/7-6/18/94). Baseline cultures as follows: 2 out of 2 blood cultures grew K. pneumoniae. Pertinent Med Hx included: peripheral edema, HTN, tachycardia, gallstones, ascites, anorexia, bloody emesis, bloody diarrhea, hematuria, hypothyroid, hyponatremia, coagulopathy, pancytopenia, tremors, anxiety, depression, hallucinations, decubitus ulcers, oral herpes simplex infection, and oral candidiasis. Pertinent concomitant therapy included: nystatin, acyclovir, silvadene, Lasix, Pepcid, Synthroid, benadryl, Tylenol, Atropine, G-CSF, Vit K, Dilaudid, urokinase, Phenergan, PRBCs, platelets, and electrolyte supplementation. Pertinent laboratory values: platelet count=26,000 at baseline (6/7), and 32,000 on 6/22/94. The patient defervesced, neutropenia resolved, repeat blood cultures were negative, and study therapy was stopped on 6/18. However, her overall status declined during the hospitalization and the family elected for supportive care only on 6/23. The patient died on (b) (6) days post therapy). No autopsy was performed. The investigator deemed that the patient’s death was caused by progression of metastatic breast cancer. The following SAEs were noted in the CRF: mild vaginal bleeding (6/20 to death) mild skin breakdown of thighs and perineum (6/20 to death), and progression of breast cancer (7/1 to death). None of these SAEs were considered related to study therapy.

M.O. comment: This patient’s death was most likely due to progression of widely metastatic breast cancer.

Patient AI411-204-010-189 was an 81 yo Black female from the USA with immunoblastic lymphoma. Recent chemotherapy (5/27/94 only) included: Cytoxan. On 6/14/94, she developed febrile neutropenia (Temp=102.4 F, ANC=270) and on 6/15/94 received ceftazidime 2gIVq8h for 5 days (6/15-6/19/94). Baseline blood and urine cultures were negative. Pertinent Med Hx included: malnutrition, ulcer, anorexia, pancytopenia, elevated hepatic enzymes, diarrhea, hematuria, and intermittent disorientation. Pertinent concomitant therapy included: Tagamet, Prilosec, Pepcid, Tylenol, Phenergan, G-CSF, Dilaudid, PRBCs, and platelets. Pertinent laboratory values: hemoglobin=6.9 (6/14) and 9.5 (6/23); platelets =55,000 (6/14) and 37,000 (6/23); creatinine=0.7 (6/14) and 1.0 (6/23). Repeat blood cultures were negative. The patient completed therapy on 6/19 and was discharged on (b) (6) On(b) (6) she was re-admitted with decreased level of consciousness and lack of oral intake. She had a fever=100.8 F, blood cultures were negative, and was empirically placed on ampicillin. CT scan showed progression of lymphoma. On 6/24, she was made comfort care only. The patient died on (b) (6) days post therapy). No autopsy was performed. The investigator deemed that the patient’s death was caused by lymphoma. The following SAEs were noted in the CRF: slurred speech 6/20 to

129

death), dehydration (6/22-6/24), and progression of lymphoma (6/19 to death). None of these SAEs were considered related to study therapy.

M.O. comment: This patient’s death was most likely due to progression of lymphoma.

Patient AI411-204-010-369 was a 53yo White male from the USA with chronic myelogenous leukemia (CML) in blast crisis. Recent chemotherapy (9/16-9/18/93) included: vincristine, idarubicin, and methotrexate. On 9/26/93, he developed febrile neutropenia (Temp=38.8 C, ANC<100) and received ceftazidime 2gIVq8h for 4 days (9/26-9/29/93). Baseline blood and urine cultures were negative. S. pneumoniae was isolated from sputum. Pertinent Med Hx included: hyperglycemia, dysphagia, oral and dermal candidiasis, right sided facial droop, positive tobacco use, and Ativan allergy (unknown reaction). Pertinent concomitant therapy included: fluconazole, acyclovir, vancomycin (9/28 to death), ciprofloxacin (9/22-9/26), ceftazidime (9/29-10/2), amphotericin B (10/1 to death), Flagyl (10/4 to death), insulin, aldactone, PRBCs, platelets, TPN and Intralipid. Pertinent laboratory values: creatinine=0.9 (9/26) and 1.2 (10/4). At the time of enrollment, the patient was concomitantly treated with fluconazole for oral and dermal candidiasis. During study therapy, the patient had persistent fevers despite negative follow-up sputum cultures on 9/30 and 10/3. On 9/28, vancomycin was added and on 9/29 the patient was placed on non-study ceftazidime. On 9/28 and 10/3, C. krusei was isolated from a blood culture and on 10/7 from a Hickman catheter. Amphotericin B was started on 10/1. The patient continued to deteriorate and died on (b) (6) days post therapy). No autopsy was performed. The investigator deemed that the patient’s death was caused by leukemia in blast crisis and sepsis and not related to study therapy. The following SAEs were noted in the CRF: lethargy (10/2 to death), shortness of breath (10/3 to death), and continued fever and sepsis (9/26 to death). None of these SAEs were considered related to study therapy.

M.O. comment: This patient’s death was most likely due C. krusei sepsis and leukemia in blast crisis.

Patient AI411-204-013-161 was a 62 yo Black female from the USA with metastatic breast cancer (metastases to bone and lung). Recent chemotherapy (2/1/94 only) included: Cytoxan, methotrexate, 5-FU. On 2/7/94, she developed febrile neutropenia (Temp=103.5 F, ANC=464) and on 2/8/94 received ceftazidime 2gIVq8h for 12 days (2/8-(b) (6) Baseline cultures were as follows: urine culture grew E. coli on 2/8, blood cultures were negative. Pertinent Med Hx included: pleural effusion requiring chest tube, pancytopenia, mucositis, and s/p hemiarthroplasty 2/93. Pertinent concomitant therapy included: fluconazole, Monistat vaginal suppository, morphine, leucovorin, scopolamine, Tylenol, Phenergan, lasix, tamoxifen, verapamil, clonidine patch, electrolyte supplementation, nifedipine, albuterol, bleomycin sclerosis of pleura, platelets and PRBCs. The subject defervesced, repeat urine culture on 2/13 was negative, and study therapy was discontinued on (b) (6) Seventeen hours post the last dose of ceftazidime, the patient was found unresponsive and in cardiac arrest without blood pressure, pulse or respirations. No resuscitative efforts were made and the patient died (b) (6) No autopsy was performed. The investigator deemed that the patient’s death was caused by metastatic breast cancer and unrelated to study therapy. The following SAE was noted in the CRF: cardiovascular collapse (b) (6) -death) deemed unrelated to study drug.

130

M.O. comment: This patient’s death was most likely due cardiorespiratory arrest associated with metastatic breast cancer. Though less likely, the M.O. cannot completely rule out the possibility that an unforeseen toxic effect from the study drug may have contributed to the patient’s death given that it occurred within hours of the last dose of study therapy.

Patient AI411-204-016-406 was a 61 yo White male from the USA with acute myelomonocytic leukemia and myelodysplastic syndrome. Recent chemotherapy (9/20-9/27/93) included: Ara-C and daunomycin. On 10/6/93, he developed febrile neutropenia (Temp=39.3 C, ANC<100) associated with a suspected pneumonia and received one dose of ceftazidime 2gIV (10/6/93). Baseline cultures were as follows. Urine culture was negative. Blood cultures grew Micrococcus species in 4 of 6 blood cultures. Sputum cultures were not found in the CRF. Pertinent Med Hx included: leukemia cutis, HTN, hypokalemia, hypoglycemia, pulmonary edema post transfusion on 9/29/93. Pertinent concomitant therapy included: antibacterial agents as discussed below, fluconazole, norfloxacin, INH, rifampin, Tylenol, benadryl, lasix, Robitussin cough syrup, hydrocortisone, labetolol, Lomotil, electrolyte supplementation, tranexamic acid, Demerol, morphine, verapamil, vitamin K, albumin, Solumedrol, digoxin, versed, carafate, thiamine, hydralazine, vasotec, enalapril, adenosine, dopamine, platelets, PRBCs, and fresh frozen plasma. The CRF contained lab values for only 10/6: hemoglobin=5.0, platelets=21,000 and creatinine=1.0. At the time of enrollment, the patient was hypotensive (SBP=84), tachypneic (RR=40), and tachycardic (HR=150). CXR revealed a bilateral, multilobar process. The patient received one dose of study medication and then the treating physician changed therapy to vancomycin, amikacin, imipenem, and erythromycin on 10/6. Despite aggressive ICU (b) (6) (b) management, the patient died on (post therapy day (6) . No autopsy was performed. The investigator deemed that the patient’s death was caused by acute respiratory failure, pulmonary edema, and pulmonary hemorrhage due to acute leukemia and myelodysplastic syndrome, and not related to study therapy. The following SAEs were noted in the CRF: respiratory failure (10/6 to death), pulmonary hemorrhage (10/6 to death), and GI bleeding (10/7 to death). None were deemed related to study therapy.

M.O. comment: This patient’s death was most likely due acute respiratory failure and pulmonary hemorrhage associated with acute leukemia and myelodysplastic syndrome.

Patient AI411-204-016-484 [a.k.a. AI411-204-016-401] was a 66 yo White female with chronic lymphocytic leukemia (CLL). [The patient had successfully completed a 7 day course of cefepime 2gIVq8h (9/4-9/10/93) for a prior episode of febrile neutropenia in the same study.] While on pulse decadron maintenance therapy for her CLL, the patient developed febrile neutropenia (Temp=38.9 C, ANC=0) and received 4 days of ceftazidime 2g (2/13-2/16/94). Baseline blood cultures were negative, urine cultures contained mixed flora. Med Hx included: low-grade Hodgkin’s lymphoma, COPD, pancytopenia, cranial nerve 7 palsy, sacral decubitus ulcer, and s/p hysterectomy. Pertinent concomitant therapy included: Tylenol, Robitussin, Percocet, Narcan, Lasix, dopamine, PRBCs, platelets, and electrolyte supplementation. During the hospital course, the patient reportedly developed decreased level of consciousness and lethargy that progressed to unresponsiveness, edema, and decreased urine output. The patient died on (b) (6) . The investigator deemed that the patient’s death was caused by CLL and low- grade Hodgkin’s lymphoma. No autopsy was performed. The following SAE was noted in the

131

CRF: lethargy (2/16/94 to death), edema (2/17 to death), decreased urine output (2/17 to death), and unresponsiveness (2/17 to death).

M.O. comment: This patient’s death was most likely due to underlying chronic lymphocytic leukemia and Hodgkin’s lymphoma.

Patient AI411-204-016-485 was a 31 yo Black male from the USA with mycosis fungoides and high grade immunoblastic T-cell lymphoma. Recent chemotherapy (2/25-2/28/94) included: VP 16, vincristine, and Adriamycin. On 3/7/94, he developed febrile neutropenia (Temp=38.2 C, ANC=100) and received ceftazidime 2gIVq8h for 4 days (3/7-3/10/94). At baseline, he had tachypnea, tachycardia, cough, mucositis, and multiple open skin lesions (related to the mycosis fungoides), one of which had serosanguinous drainage. Baseline cultures results were as follows: 2 out of 4 wound cultures were positive for S. aureus (MSSA); urine culture was positive for MSSA, 2 out of 2 oral lesion cultures grew HSV 1, blood cultures were negative, and sputum cultures grew normal flora. Pertinent Med Hx included: pancytopenia, hematuria, hyponatremia, hypokalemia, splenomegaly, anxiety, and tobacco and alcohol abuse. Pertinent concomitant therapy included: antibacterial agents as discussed below, fluconazole, bactrim three times per week, silvadene cream, Tylenol, Ativan, MS contin, vitamin K, electrolyte supplementation, heparin, total parenteral nutrition, PRBCs, and platelets. Pertinent laboratory values: creatinine=0.4 on 3/7 and 0.4 on 3/10 (last recorded in CRF). Due to persistent fevers, study therapy was discontinued on 3/10, and replaced with nafcillin, ceftazidime (non-study), and acyclovir (for herpes simplex infection). Fevers persisted. On 3/12, the patient was placed on a morphine drip for pain. On 3/13, he was diagnosed with renal dysfunction associated with vomiting, dyspnea, and abnormal renal lab values. Lethargy increased and on 3/15, the patient became hypothermic (36.8 C). On 3/16, all medications except morphine were discontinued. (b) (6) ( The patient died on (b days post therapy). No autopsy was performed. The investigator deemed that the patient’s death was caused by infection, renal failure, and high grade lymphoma and unrelated to study therapy. The following SAEs were noted in the CRF: hypothermia (3/16 to death) and renal insufficiency (3/14 to death); both deemed unrelated to study therapy.

M.O. comment: This patient’s death was most likely due to S. aureus (MSSA) sepsis, renal failure, and high grade lymphoma. The M.O. cannot rule out the possibility that the patient’s death was related to study drug failure from ceftazidime. Ceftazidime may have also caused or contributed to the patient’s renal failure.

Patient AI411-204-017-424 was a 27 yo Black female from the USA with acute myelogenous leukemia. Recent chemotherapy (5/31-6/15/94) included: idarubicin and cytosine arabinoside followed by cytarabine (6/16-6/18) and mitoxantrone (6/20-6/22) while on study therapy. On 6/16/94, she developed febrile neutropenia (Temp=40.3 C, ANC<100) and received ceftazidime 2gIVq8h for 7 days (6/16-6/22/94). At baseline, she had thrombocytopenia (platelets=800) and significant vaginal bleeding. Baseline blood and urine cultures were negative. Pertinent Med Hx included: pancytopenia. Pertinent concomitant therapy included: gentamicin (6/18-6/19), Flagyl (6/16-6/22), piperacillin (6/18-6/19), vancomycin (6/19-6/22, 6/27 to death), ceftazidime 6/25 to death), codeine, benadryl, Tylenol, allopurinol, ondansetron, prochlorperazine, diazepam, hydrocortisone, electrolyte supplementation, Lasix, loperamide, meperidine, and nifedipine. On 6/20, the patient experienced unsteady gait, slurred speech, and dizziness thought secondary to

132

Ara-C cerebellar toxicity. On 6/22, study drug was discontinued and switched to vancomycin and non-study ceftazidime. Marrow failed to recover and vaginal bleeding worsened. On (b) (6) days post therapy), the patient died. No autopsy was performed. The investigator deemed that the patient’s death was caused by progressive leukemia and unrelated to study therapy.

M.O. comment: This patient’s death was most likely due progressive leukemia.

Study AI411-143 Non-Comparative Study

The CRF for the following patient was not provided by BMS because the patient died >30 days after the last dose of cefepime. The following was based on information found in the Death Narrative. Patient AI411-143-001-011 was a 60 yo male (race unknown, country unknown) with Waldenstrom’s disease. Recent chemotherapy (dates not provided) included: chlorambucil. On 4/13/90, he developed febrile neutropenia (Temp=not provided, ANC=378) and received cefepime 2gIVq8h for 11 days (4/14-4/24). Baseline cultures were negative. Pertinent Med Hx included: tuberculosis adenitis resected in 1949. Concomitant therapy was not provided. Laboratory values were not provided. After 11 days of cefepime therapy, the patient’s baseline cough had improved, however fever and chills remained. Cefepime was discontinued. The patient died on (b) (6) days post the last dose of cefepime. Death was reportedly due to “consequences of the underlying disease and a massive lung edema, as was confirmed by the autopsy”. The investigator deemed that the patient’s death was unrelated to cefepime administration. No SAEs were reported in the narrative.

M.O. comment: Based on BMS’s Death Narrative, this patient’s death was due to his underlying Waldenstrom’s disease and pulmonary edema. This patient died > 30 days after the last dose of cefepime.

Patient AI411-143-001-029 was a 63 yo White male from Switzerland with metastatic lung cancer (metastases to liver and pancreas). Recent chemotherapy (10/17-10/24/90) included: cisplatin, mitomycin, and vindesine. On 10/28, he developed febrile neutropenia (Temp=38.3 C, ANC<20) and received cefepime 2gIVq8h for 6 days (10/29 (b) (6) Baseline blood cultures x2 were positive for S. pneumoniae. CXR showed a left upper lobe pneumonia. Pertinent Med Hx included: atrial fibrillation, right pneumonectomy 10 months prior to study entry, biliary prosthesis due to metastatic hepatic involvement. Pertinent concomitant therapy included: PRBCs, acyclovir, digitalis, lasix, morphine, lactulose, and electrolyte supplementation. Pertinent laboratory values included: creatinine=101 (units?) on 10/28 and 94 (units?) on 11/1. After 5 days of therapy, the patient developed ARDS and was placed on comfort measures. The patient went into a coma and died the following day on (b) (6) No autopsy was performed. The investigator deemed that the patient’s death was due to ARDS due to extensive pneumonia, congestive heart failure, and the underlying metastatic lung cancer. The following SAE was noted in the CRF: “death” (b) (6) ) was deemed unrelated to study therapy.

M.O. comment: This patient’s death was most likely due to S. pneumoniae pneumonia and bacteremia complicated by ARDS, CHF, and metastatic lung cancer.

133

The following patient died > 30 days post study therapy. Patient AI411-143-001-036 (also enrolled as Patient AI411-143-001-013 for a prior episode of febrile neutropenia) was a 31 yo White male from Switzerland with relapsed leukemia s/p two autologous bone marrow transplants. Recent chemotherapy (10/23-10/27/90) included: mitoxantrone and etoposide. On 10/28, he developed febrile neutropenia (Temp=38.4 C, ANC=565) with an inflamed catheter site and received cefepime 2gIVq8h for 3 days (10/28-10/30). Baseline catheter and catheter site cultures grew S. epidermidis resistant to cefepime. Baseline blood and urine cultures were negative. Pertinent Med Hx was included: candidal pharyngitis. Pertinent concomitant therapy included: norfloxacin, PRBCs, platelets, fluconazole, acyclovir, prednisone, and allopurinol. On Day 3, due to persistent fever and catheter site inflammation, cefepime was discontinued and the patient was placed on vancomycin and imipenem. The patient’s leukemia did not respond to chemotherapy and he died on (b) (6) days post cefepime). Autopsy status was unknown per the CRF. The investigator deemed that the patient’s death was due to refractory leukemia. No SAEs were noted in the CRF.

M.O. comment: This patient’s death was most likely due to refractory leukemia.

Patient AI411-143-001-068 (also enrolled as Patient AI411-143-001-059 for a prior episode of febrile neutropenia) was a 73 yo White male from Switzerland with acute non-lymphoblastic lymphoma. Recent chemotherapy (5/21-5/25/91) included: amsacrine and vepeside. On 5/30, he developed febrile neutropenia (Temp=40.0 C, ANC<200) and received cefepime 2gIVq8h for 8 days (5/30-6/6). Baseline blood cultures were positive for Stomatococcus sp., urine culture was negative. CXR showed a massive left sided effusion. Pertinent Med Hx included: CHF, HTN, and a pleural effusion related to CHF and tumor infiltration. Pertinent concomitant therapy included: ciprofloxacin, penicillin, ranitidine, digoxin, allopurinol, prednisone, triazolam, electrolyte supplementation, PRBCs, and platelets. The patient’s fever initially resolved; however on 6/6 (Day 8) the patient developed a new fever and was diagnosed with Candida albicans fungemia. The patient died on (b) (6) . Autopsy revealed pulmonary emboli and disseminated candidiasis (cultures positive from lungs, liver, and spleen). The investigator deemed that the patient’s death was due to disseminated candidiasis and pulmonary emboli, with CHF, massive left pleural effusion, and acute leukemia as contributing factors. The following SAE was noted in the CRF: “death” (b) (6) was deemed unrelated to study therapy.

M.O. comment: This patient’s death was most likely due to disseminated candidiasis and pulmonary emboli (which may have been a manifestation of the candidiasis). CHF, massive left pleural effusion, and acute leukemia probably contributed to the patient’s death.

Patient AI411-143-002-005 was a 60 yo White male from Belgium with myelodysplasia with refractory anemia. No recent chemotherapy was noted. On 3/30/90, he developed febrile neutropenia (Temp=38.0 C, ANC=483) and received cefepime 2gIVq8h for 19 days (3/30 4/17/90). Culture results below. Pertinent Med Hx included: peripheral vascular disease, anthrax 30 yrs prior, petechial lesions on legs, urinary incontinence, fatigue, depression, 2.5 ppd tobacco x 40 yrs, and 2 beers/day. Pertinent concomitant therapy included: antimicrobials as noted below, leucovorin, itraconazole, and vitamin B6. On 4/2 a mouth culture was positive for C. albicans, on 4/3 H. simplex was cultured from a mouth lesion and Corynebacterium sp. from a

134

blood culture. On 4/4, M. tuberculosis was isolated from a bronchial culture aspirate. The patient was diagnosed with pulmonary and urinary tuberculosis. Over the course of the hospitalization, rifampicin, isoniazid, ethambutol, amphotericin, bactrim, acyclovir, nystatin, and vancomycin were added to the patient’s cefepime therapy. On 4/17, the patient had defervesced and cefepime was discontinued. On 4/27, the patient developed a new fever that was thought to be due to PRBC/platelet transfusions and the patient was sent home the following day. On (b) (6) days post therapy), the patient died at home. No autopsy was performed. The investigator deemed that the patient’s death was caused by “the consequences of myelodysplasia, tuberculosis and the episode of fever and neutropenia”, and was not related to cefepime. The following SAE was noted in the CRF: “death” (b) (6) was deemed unrelated to study therapy.

M.O. comment: There was not enough information in the CRF or Death Narrative to (b) determine what acutely caused this patient to die at home(6) days post cefepime; however, ultimately, the patient likely died from disseminated tuberculosis and pancytopenia associated with myelodysplasia.

Patient AI411-143-002-009 was a 58 yo Asian male from Belgium with metastatic seminoma (metastases to the aorta). Recent chemotherapy (3/2-3/28/90) included: etoposide, cisplatin, and bleomycin. On 4/14/90, he developed febrile neutropenia (Temp=38.0 C, ANC=390) and received cefepime 2gIVq8h for 7 days (4/15-4/21). Baseline blood cultures x2 were positive for S. epidermidis. CXR from 3/22 was reported as negative, and CXR from 4/18 demonstrated bilateral pulmonary infiltrates vs. pneumonia. Pertinent Med Hx included: left orchiectomy in 1984, prostatectomy in 1988, and s/p laparotomy 1/90 to evaluate/debulk retroperitoneal metastases. Pertinent concomitant therapy included: tilidine/naloxone, paracetamol, nystatin, and acetylcystine. The patient initially defervesced; however fever, chills and lethargy returned on 4/16. A repeat blood culture on 4/17 was negative; however, the patient manifested clinical signs of pulmonary infiltrative disease vs. pneumonia which was confirmed on CXR (4/18). Vancomycin was started on 4/17 with a reportedly a good response. Cefepime was stopped on 4/21. Additional antimicrobial therapy during the course of the hospitalization included: pefloxacin, Bactrim, and ceftazidime. The patient’s condition continued to deteriorate and he was placed on comfort measures on (b) (6) and died on(b) (6) days post cefepime). No autopsy was performed. The investigator deemed that the patient’s death was due to metastatic seminoma, interstitial pneumonia, and multiple organ failure. No SAEs were noted in the CRF.

M.O. comment: This patient’s death was most likely due to metastatic seminoma, interstitial pulmonary process (bleomycin is known to cause pulmonary fibrosis) versus pneumonia (pathogen unknown), and multiple organ failure.

Patient AI411-143-002-017 was an 84 yo White female from Belgium with myelodysplasia in blast crisis. Chemotherapy provided during the study included etoposide (5/19-5/20/90). On 5/14, she developed fever, but was not neutropenic (Temp=39.2 C, ANC=10,860) and received cefepime 2gIVq8h for 10 days (5/14-5/23). Baseline blood and urine cultures were negative. Culture of a right hip hematoma/abscess grew S. epidermidis. CXR was negative. Pertinent Med Hx included: “urinary crystalia”, osteoarthritis, lower extremity edema, tachycardia, and CHF with orthopnea. Pertinent concomitant therapy included: PRBCs, platelets, isosorbide dinitrate, lorazepam, clindamycin (5/17-5/23), furosemide, KCL, nystatin, and morphine.

135

Pertinent laboratory values included: creatinine=0.9 on 5/15 and 3.9 on 5/23. Clindamycin was added on 5/17 (Day 3) due to persistence of fever, dyspnea and worsening right hip lesion. The patient was diagnosed with myelodysplasia in blast crisis and she began to deteriorate with multisystem organ failure: cardiac, renal, hepatic, mental status, and respiratory failure. On 5/22, the patient experienced a hypotensive episode and on 5/23 was placed on comfort measures. The patient died on (b) (6) post cefepime). Autopsy “confirmed the hematologic, pulmonary and hepatic clinical status”. The investigator deemed that the patient’s death was due to myelodysplasia and the suspension of all therapy (comfort measures only). The following SAE was noted in the CRF: “death” (b) (6) was deemed unrelated to study therapy.

M.O. comment: This patient’s death was most likely due to myelodysplasia in blast crisis accompanied by progressive multiple organ failure.

Patient AI411-143-002-021 (also enrolled as Patient AI411-143-002-015 for a prior episode of febrile neutropenia) was an 18 yo Asian male from Belgium with acute lymphocytic leukemia s/p autologous bone marrow transplant (3/25/90). Recent chemotherapy (5/24/90) included: vincristine; he also received radiotherapy (3/19-3/24/90). On 6/1/90, he developed febrile neutropenia (Temp=38.6 C, ANC=184) and received cefepime 2gIVq8h for 12 days (6/1 6/12/90). Baseline blood cultures were negative. CXR was negative. Pertinent Med Hx included: chemotherapy related hepatotoxicity, depression, asthenia, facial paralysis due to carcinomatous meningitis, oral HSV infection, and a dental abscess. Pertinent concomitant therapy included: quinolone for gut decontamination, vancomycin, Flagyl, Augmentin, amikacin, ceftazidime, PRBCs, platelets, diazepam, amitriptyline, alprazolam, itraconazole, morphine, acyclovir, electrolyte supplementation, amphotericin B, and Solucortef (with amphotericin B infusion). During the course of the study, the patient’s fever persisted so amphotericin B, itraconazole, vancomycin, and acyclovir were added between Days 3-5. On 6/6, the patient manifested HSV infection of the moth and eye. The patient was afebrile on 6/11-6/12 and cefepime was stopped. The patient reportedly remained afebrile until 6/24 when he developed a new fever. Blood cultures from 6/25 and 6/26 were positive for P. aeruginosa and received Augmentin, amikacin and ceftazidime. On 6/27, the patient was placed on palliative care and died on(b) (6) days post therapy). No autopsy was performed. The investigator deemed that the patient’s death was due to P. aeruginosa septicemia and acute leukemia. No SAEs were noted in the CRF.

M.O. comment: This patient’s death was most likely due to P. aeruginosa sepsis and acute leukemia. It is unlikely that the patient’s death was due to study drug failure given that the patient was afebrile for approximately 12 days between the end of cefepime therapy and the start of the new fever.

Patient AI411-143-002-038 was a 44 yo White female from Belgium with metastatic breast cancer (metastases to lungs, liver, brain, and kidneys). She underwent recent radiation therapy from 7/18-7/31/90 and chemotherapy on 8/2 only: cisplatin, Adriamycin, and cyclophosphamide. On(b) (6) she developed febrile neutropenia (Temp=”>38.0 C”, ANC not performed; however WBC=0.15) with diarrhea, melena, chills, and received a single dose of cefepime 2gIV (8/9/90 only). Baseline blood culture yielded C. septicum which was resistant to cefepime. Pertinent Med Hx included: gastro-duodenal ulcer. Pertinent concomitant therapy included:

136

hydrocortisone, paracetamol, and potassium supplementation. The patient rapidly deteriorated into septic shock post enrollment and died within hours of the 1st dose of cefepime. No autopsy was performed. The investigator deemed that the patient’s death was caused by C. septicum septic shock and metastatic cancer. The following SAE was noted: death due to septic shock.

M.O. comment: This patient’s death was most likely due to C. septicum septic shock associated with widely metastatic cancer. This pathogen was resistant to cefepime.

Patient AI411-143-002-040 was a 58 yo White female from Belgium with metastatic breast cancer (pulmonary and bone metastases). She underwent radiation therapy on 7/16/90. On (b) (6) she developed febrile neutropenia (Temp=38.9 C, ANC=0) and received a single dose of cefepime 2gIV (b) (6) only). Baseline cultures as follows: blood culture positive for methicillin-susceptible S. aureus (MSSA); left breast wound culture positive for MSSA, E. cloacae, Peptostreptococcus spp., Porphyromonas sp., Bacteroides sp.; urine culture positive for MSSA and E. coli; vaginal swab positive for E. coli; and sputum positive for K. pneumoniae. Pertinent Med Hx included: skin ulcer, hepatic steatosis, alcoholism, “psychologic problems”, polytrauma at age 30 years. Pertinent concomitant therapy included: clindamycin ( (b) (6) only), midazolam, etomidate, adrenaline, Levophed, dopamine, dobutamine, PRBCs, platelets, albumin The patient was enrolled with sepsis and a systolic blood pressure=95. Thirty minutes post enrollment and, “while the 2 gram cefepime infusion set was set up, the patient became comatose and died shortly after.” The patient sustained cardiorespiratory arrest and died the same day (b) (6) ). Autopsy revealed breast cancer with pulmonary congestion, “gangliar and hepatic metastases”, and fungal spores and filaments in the lungs. The investigator deemed that the patient’s death was caused by septic shock and underlying metastatic cancer and was unrelated to cefepime. The following SAE was noted in the CRF: “death due to septic shock” was deemed unrelated to cefepime.

M.O. comment: This patient’s death was most likely due to MSSA septic shock and underlying widely metastatic breast cancer. The M.O. cannot rule out cefepime treatment failure; however, the M.O. also notes that the patient was enrolled with sepsis and apparently progressed to septic shock prior to completion of the first and only infusion of cefepime.

Study AI411-158 Non-Comparative Study

Patient AI411-158-001-001 was a 46 yo White male from the Netherlands with chronic myeloid leukemia, s/p bone marrow transplant (1/91). Recent chemotherapy (1/24-1/25/91) included: cyclophosphamide; he also received radiotherapy (1/20-1/29/91). On 2/3, he developed febrile neutropenia (Temp=39.8 C, ANC<100) and received cefepime 2gIVq8h for 12 days (2/3 2/14/91). Baseline blood and urine cultures were negative. CXR was negative. No additional pertinent Med Hx was provided in the CRF. Pertinent concomitant therapy included: colistin and ciprofloxacin for gut decontamination and amphotericin B for antifungal prophylaxis; teicoplanin, cyclosporine (1/20-2/17), fluconazole, lasix, codeine, itraconazole, and Zovirax. The patient completed 12 days of cefepime and teicoplanin. Then on 2/13, 2/14, 2/16, and 2/19, blood cultures were positive for C. krusei. Around the same time, the patient’s hepatic function began to deteriorate. On 2/15, ciprofloxacin and amphotericin B were started. The patient

137

continued to deteriorate and die on (b) (6) days post cefepime). No autopsy was performed. The investigator deemed that the patient’s death was due to disseminated candidiasis due to C. krusei, and progressive liver and respiratory failure. The investigator felt that the hepatic failure could have been due to the candidiasis, veno-occlusive disease related to a mismatched bone marrow transplant, or toxin. No SAEs were noted in the CRF.

M.O. comment: This patient’s death was most likely due to disseminated candidiasis by C. krusei. Hepatic failure may have been due to disseminated candidiasis, veno-occlusive disease (graft versus host disease) related to mismatched bone marrow transplant, or toxic agent (including cefepime, other concomitant meds, or chemotherapy).

Patient AI411-158-001-011 was a 43 yo White female from the Netherlands with chronic myeloid leukemia, s/p bone marrow transplant (4/91). Recent chemotherapy included: cyclophosphamide (4/11-4/12/91) and busulfan (11/16/90-4/14/91); she also received radiotherapy (4/15-4/16/91). On 4/21, she developed febrile neutropenia (Temp=39.4 C, ANC<100) and received cefepime 2gIVq8h for 10 days (4/21-4/30). Baseline blood culture grew Streptococcus sanguis. CXR was negative. Pertinent Med Hx included: recurrent erysipelas. Pertinent concomitant therapy included: ciprofloxacin for gut decontamination and amphotericin B for antifungal prophylaxis; insulin, lasix, electrolyte supplementation, Mesna, Zovirax, bactrim twice weekly, itraconazole, heparin, prednisone, cyclosporine, and Soldactone. Pertinent laboratory studies included: ALT=35 on 4/22, 118 on 4/25, and 295 on 5/2; total bilirubin=56 on 4/22, 104 on 4/25, and 640 on 5/2; and creatinine=74 on 4/22, 80 on 4/25, and 280 on 5/2. The patient became afebrile by 4/25 and her ANC rose to 800 by 4/30. On 4/30, cefepime was stopped, and ciprofloxacin was started (reason unknown, possibly as prophylaxis against spontaneous bacterial peritonitis given that her hepatic function was deteriorating). The patient died on (b) (6) days post cefepime). Autopsy reportedly showed “ascites and pleural effusion with enlarged lines possibly due to veno-occlusive (disease)”. The investigator deemed that the patient’s death was due to hepatic failure from veno-occlusive disease (graft versus host disease). The investigator noted that the veno-occlusive disease was probably related to busulfan use; however, “It cannot be 100% excluded that cefepime contributed to the liver disease although it seems unlikely.” No SAEs were noted in the CRF.

M.O. comment: This patient’s death was most likely due to hepatic failure related to veno- occlusive disease (graft versus host disease) associated with busulfan use. The M.O. cannot completely rule out the possibility that cefepime may have contributed to the patient’s hepatic failure.

Patient AI411-158-001-017 was a 44 yo White female from the Netherlands with acute myelogenous leukemia, s/p bone marrow transplant (5/15/91). Recent chemotherapy (5/9 5/10/91) included: cyclophosphamide; she also received radiotherapy (7/10-7/12). On 7/3, she developed febrile neutropenia (Temp=38.4 C, ANC<100) and received cefepime 2gIVq8h for 9+6=15 days total (7/3-7/11 and 7/15-7/20). Baseline blood culture yielded methicillin-resistant S. epidermidis (MRSE) that was thought to be a contaminant. CXR was negative. No additional pertinent Med Hx was included in the CRF. Pertinent concomitant therapy included: Bactrim twice weekly, acyclovir, Orgametril, cyclosporine, ondansetron, lasix, electrolyte supplementation, prednisolone, morphine, oxazepam, insulin, fluconazole, and PRBCs.

138

Pertinent laboratory studies included: creatinine=133 on 7/4, 77 on 7/7, 115 on 7/16, 233 on 7/19, and 447 on 7/21. The patient initially defervesced by 7/5, repeat cultures were negative, and cefepime was discontinued on 7/11. However, on 7/15, the patient developed another fever and she had pain and erythema around her catheter site. Cefepime in combination with teicoplanin was restarted. Around the same time, the patient began developing renal failure which was thought to be secondary to cyclosporine toxicity. The patient underwent a second bone marrow transplant on 7/17/91. The patient continued to deteriorate and the decision was made to place the patient on comfort care. She died on (b) (6) days post therapy). Autopsy showed bone marrow hypoplasia, pulmonary hemorrhage and infarction, and thrombosis around the central venous catheter. The investigator deemed that the patient’s death was due to renal failure related to cyclosporine toxicity and long-term bone marrow aplasia. The investigator also suspected bronchopneumonia. No SAEs were noted in the CRF.

M.O. comment: This patient’s death was due to multiple organ failure (renal failure, bone marrow failure, and pulmonary hemorrhage resulting in respiratory failure). The renal failure may have been related to cyclosporine toxicity, other concomitant medications, leukemia, and/or cefepime. Bone marrow failure may have been due to bone marrow transplant failure, leukemia, and/or cefepime. Pulmonary hemorrhage may have been related to chronic thrombocytopenia, barotrauma, leukemia, and/or bronchopneumonia that may have been due to a bacterial, viral, fungal, or opportunistic pathogen.

Patient AI411-158-001-025 was a 23 yo White female from the Netherlands with chronic myeloid leukemia s/p bone marrow transplant on 9/14/91. The patient received radiotherapy (8/26/-8/27/91) and chemotherapy (8/15-8/23/91) with idarubicin, cytarabine, and cyclophosphamide. On 9/24/91, she developed febrile neutropenia (Temp=39.2 C, ANC < 100) and received cefepime 2gIVq8h for 16 + 2=18 days total (9/24-10/8 and 10/10-10/11). At presentation, the patient was jaundiced (total bilirubin=498). Baseline blood, urine, and sputum cultures were negative. CXR was negative. No additional pertinent Med Hx was included in the CRF. Pertinent concomitant therapy included: ciprofloxacin 1 g/day (9/19-9/26/91), amphotericin B (8/15-10/12), clotrimazole twice weekly, acyclovir, itraconazole, ondansetron, Lasix, hydrocortisone, morphine, famotidine, insulin, penicillin G (8/22-9/4/91), GM-CSF, parenteral nutrition, and “blood products”. Pertinent laboratory results included: hemoglobin on 9/25=6.8 and on 10/10=6.0, platelet count on 9/25=18,000 and on 10/10=29,000. Throughout the hospital course, the patient remained febrile and neutropenic. Repeat blood cultures from 9/29-10/7 were negative. Teicoplanin was added on 10/3. On 10/8, the patient developed a skin rash of moderate severity that was considered probably related to cefepime, and both cefepime and teicoplanin were discontinued on the same day. Ciprofloxacin was prescribed from 10/9 10/10, followed by cefepime rechallenged from 10/10-10/11. Skin rash reportedly did not improve off cefepime or teicoplanin. The patient continued to deteriorate and she died on (b) (6) post therapy). Autopsy revealed bone marrow hypoplasia, hemorrhagic cystitis, and pulmonary edema. The investigator deemed that the patient’s death was caused by hemorrhagic cystitis and persistent aplasia post bone marrow transplant. The following AE was noted in the CRF: moderate severity skin rash (10/8 to death) and deemed probably related to cefepime; however did not improve off cefepime.

139

M.O. comment: This patient’s death was most likely due to hemorrhagic cystitis and long term aplasia s/p bone marrow transplant.

The CRF for the following patient was not provided by BMS because the patient died >30 days after the last dose of cefepime. The following was based on information found in the Death Narrative. Patient AI411-158-001-027/BB was a 53 yo White male (country unknown) with chronic myeloid leukemia s/p bone marrow transplant 10/2/91. Recent chemotherapy not provided. On 10/7, he was diagnosed with E. coli and S. sanguis bacteremia and received cefepime (dose not provided) for 15 days (10/7-10/21). CXR was negative. No additional pertinent Med Hx was provided. No additional concomitant therapy was provided. On 10/21, cefepime was discontinued and IV amphotericin B was started to treat suspected pulmonary aspergillosis. The infection did not respond to therapy and the patient died on (b) (6) days post cefepime). Autopsy status was not provided.

M.O. comment: Based on BMS’s Death Narrative, this patient’s death was due to suspected pulmonary aspergillosis. This patient died > 30 days after the last dose of cefepime.

4.111 COLLABORATIVE WORK WITH ANA SZARFMAN, M.D., PH.D.

WebSDM sector map

The following is a WebSDM sector map of adverse events (AEs) associated with death among the patients enrolled in the 7 BMS-sponsored comparative febrile neutropenia studies. In the following sector map, “A”= cefepime deaths associated with the listed adverse event preferred term (PT), “B” = comparator deaths associated with the listed adverse event preferred term (PT), “C” = total number of cefepime deaths, and “D” = total number of comparator deaths.

140

141

M.O. comment: In general, there were few deaths per system organ class (SOC). In the following section, the M.O. provides further analysis and discussion related to the patients that comprised the potentially biologically plausible signal SOC categories denoted in the preceding sector map.

The following are the unique subject identifying numbers for the five cefepime patients who experienced “acute respiratory distress syndrome” as a fatal AE.

USUBJID Site ID Sex Age Race Planned Arm Cefepime Plus 411198002012 002 M 50 W Vancomycin Cefepime Plus 411198002029 002 F 26 W Vancomycin Children's 411131003053 Medical Center M 1 W Cefepime 411186015004 015 M 58 Cefepime 411189010032 010 M 46 W Cefepime

The M.O. Death Narratives, Patient Profiles, and comments for the preceding five patients follow.

Patient AI411-198-002-012 [a.k.a. AI411-198-002-018] was a 50 yo White male from Belgium with acute myeloblastic leukemia. The patient had received recent chemotherapy (6/9/93); however the names of the agents were not included in the CRF. On (b) (6) , he developed febrile neutropenia (Temp=39.5 C, ANC=0) and received cefepime 2gIVq8h and vancomycin 750-2000 mg/day for 4 days ( (b) (6) ). Baseline blood cultures grew Streptococcus sanguis and Staphylococcus hemolyticus. Urine culture was negative. CXR was negative. No additional

142

pertinent Med Hx was included in the CRF. Pertinent concomitant therapy included: ciprofloxacin, fluconazole, nystatin, ondansetron, cimetidine, and Dalacin. On(b) (6) the patient developed dyspnea and tachypnea and Dalacin was added. On(b) (6) the patient developed (b) ARDS and died (on Day(6) of cefepime). No autopsy was performed. The investigator deemed that the patient’s death was due to acute myeloblastic leukemia. The following SAEs were noted in the CRF: “dyspnea/tachypnea” ((b) (6) and “ARDS” (b) (6) ) were deemed unrelated to cefepime.

M.O. comment: This patient’s death was most likely due to Streptococcus sanguis sepsis (complicated by ARDS) and acute leukemia. The M.O. cannot rule out study drug failure because the organism was susceptible to cefepime and vancomycin. Staphylococcus hemolyticus was only found in one blood culture; therefore it may have been a colonizer. Given the temporal relationship between study therapy and the onset of ARDS, the M.O. cannot completely rule out the possibility that study therapy may have been associated with the

143

ARDS; however, it is more likely that the ARDS was a sequelae of the patient’s sepsis syndrome.

Patient AI411-198-002-029 was a 26 yo White female from Belgium with Hodgkin’s lymphoma. The patient received chemotherapy (starting after enrollment, (b) (6) ; however the names of the agents were not included in the CRF. On (b) (6) she developed febrile neutropenia (Temp=39.1 C, ANC=232) and received cefepime 2gIVq8h for 6 days (b) (6) and vancomycin 1000-1500 mg/day for 5 days ( (b) (6) . Baseline blood cultures grew Stomatococcus mucilaginosus (resistant to cefepime, but susceptible to vancomycin). CXR was initially negative. No additional pertinent Med Hx was included in the CRF. Pertinent concomitant therapy included: Benzyl penicillin, ranitidine, G-CSF, morphine, epinephrine, dobutamine, dopamine, pancuronium, midazolam, and sufentanil. Repeat blood cultures from (b) (6) were negative. On (b) (6) the patient developed dyspnea and was found to have pneumonia on CXR. The patient progressed to ARDS, septic shock and multiple organ failure. (b) (6) (b) The patient died on (Day (6) of study therapy). Autopsy was performed; however results were not provided in the CRF. The investigator deemed that the patient’s death was due to pneumonia, ARDS, and septic shock. The following SAE was noted in the CRF: “ARDS” ((b) (6) (b) (6) was deemed to have an unknown relationship with study therapy.

144

M.O. comment: This patient’s death was due to multiple organ failure from pneumonia/septic shock. The pathogen isolated was Stomatococcus mucilaginosus which was resistant to cefepime. Therefore, this death did not appear to be due to failure against cefepime. Of note, the pathogen was sensitive to vancomycin. On page 37/40 of the CRF, the investigator clarified the following: “Cause of death=ARDS, resulting from infection treated in the protocol.” Given the temporal relationship between study therapy and the onset of ARDS, the M.O. cannot completely rule out the possibility that study therapy may have been associated with the ARDS; however, it is more likely that the ARDS was a sequelae of the patient’s sepsis syndrome.

145

cefepime 50mg/kgIVq8h for 2 days ( (b) (6) . Baseline blood, urine, and CSF cultures were negative. No additional pertinent Med Hx was included in the CRF. Pertinent concomitant therapy included: bactrim (b) (6) , ceftazidime (b) (6) vancomycin (b) (6) erythromycin (b) (6) ), amphotericin B ( (b) (6) , tobramycin (b) (6) , platelets, PRBCs, electrolyte supplementation, Tylenol, benadryl, morphine, and ibuprofen suspension. Pertinent laboratory values included: platelets=16,000 on(b) (6) and 11,000 on (b) (6) On the second day of therapy, the patient was found to be not eligible to continue in the study due to being < 2 years old. Therefore, cefepime was discontinued and ceftazidime was started. Over the next several days, the patient progressively deteriorated with respiratory distress and failure due to RSV pneumonia. On (b) (6) the patient died. An autopsy was performed and revealed acute myeloid leukemia, respiratory syncytial virus (RSV) pneumonia, adult-type respiratory distress syndrome, massive diffuse fungal (C. albicans) colitis, a lung tissue culture revealed C. albicans and blood cultures revealed P. aeruginosa, enterococci, and coagulase-negative Staphylococci. The investigator deemed that the patient’s death was caused by adult respiratory distress syndrome/RSV pneumonia and acute leukemia. The following SAEs were noted in the CRF: “pseudomembranous colitis” (b) (6) “ARDS” ( (b) (6) ), and “death” (b) (6) ) were deemed unrelated to cefepime.

146

M.O. comment: This patient’s death was most likely due to ARDS/RSV pneumonia. The patient’s underlying disease was acute leukemia. Given that the onset of the ARDS was over a week after the end of cefepime therapy, it is more likely that the ARDS was due to the RSV pneumonia than due to cefepime.

Patient AI411-186-015-004 was a 58 yo male (race unknown, from France) with acute non- lymphocytic leukemia. Recent chemotherapy (6/5-6/11/93) included: cytarabine and amsacrine. On (b) (6) , he developed febrile neutropenia (Temp=39.2 C, ANC<100) and received cefepime 2gIVq12h for 2 days (b) (6) ) and amikacin 400mgIVq12h for 17 days (b) (6) . One of two baseline blood cultures grew Streptococcus mitis (resistant to study therapy); urine culture was negative. CXR was negative on (b) (6) . No additional pertinent Med Hx was noted in the CRF. Pertinent concomitant therapy included: furosemide, heparin, alizapride, chlorpromazine, dexamethasone, vitamin K, paracetamol, metoclopramide, PRBCs, platelets, dobutamine,

147

dopamine, G-CSF, isosorbide dinitrate, fentanyl, midazolam, and pancuronium. The patient’s status worsened within hours of enrollment. Vancomycin was added on(b) (6) Cefepime was discontinued on (b) (6) after only 4 doses. The patient’s respiratory status progressively worsened culminating in ARDS requiring transfer to the ICU and intubation. Additional antimicrobial agents were added: piperacillin, bactrim, ganciclovir, amphotericin B, and spiramycin. (b) (6) (17 days post cefepime and 2 days post amikacin), his pulmonary status worsened, he became anuric, and died. No autopsy was performed. The investigator deemed that the patient’s death was due to S. mitis sepsis leading to acute respiratory failure. The following SAE was noted in the CRF: “ARDS” ( (b) (6) was deemed unrelated to study therapy.

M.O. comment: This patient’s death was most likely due to S. mitis sepsis leading to ARDS and multiple organ failure. The S. mitis was reported as resistant to cefepime. Given the temporal relationship between study therapy and the onset of ARDS, the M.O. cannot

148 completely rule out the possibility that study therapy may have been associated with the ARDS; however, it is more likely that the ARDS was a sequelae of the patient’s sepsis syndrome.

Patient AI411-189-010-032 was a 46 yo White male from the Netherlands with acute myeloid leukemia (AML) (date of original diagnosis not found in CRF). The CRF did not note if the patient had recently received chemotherapy for AML. He developed febrile neutropenia on (b) (6) (Temp=40.6 C, ANC=0, WBC=45.5 (88% immature forms), PLAT=11,000, Hgb=5.3) and cefepime 2 g IV (one dose on (b) (6) hours before death) was started on the same day. Baseline blood cultures x 3 were positive for Staphylococcus aureus (MSSA). Pertinent Med Hx included: HTN. Concomitant therapy included: paracetamol and fluconazole. At baseline, the patient was noted to be pancytopenic due to leukemia and thrombocytopenic. Within hours of admission, the patient developed adult respiratory distress syndrome (ARDS) secondary to pulmonary hemorrhage thought secondary to disseminated intravascular coagulation (DIC). Serious adverse events included: ARDS and cardiac arrest, and neither was deemed related to study therapy. The investigator deemed that the patient’s death on (b) (6) was due to underlying leukemia with pulmonary hemorrhage due to DIC and that the MSSA infection was not the primary cause of death.

149

M.O. comment: The patient’s death was most likely due to ARDS secondary to pulmonary hemorrhage secondary to underlying leukemia and/or MSSA sepsis. The patient died within hours of the first dose of cefepime. The M.O. cannot rule out that the patient’s death was due to study drug failure, though additional factors (overwhelming sepsis and acute leukemia) may have contributed to the patient's death. Though the M.O. cannot completely rule out the possibility that the ARDS may have been associated with cefepime exposure, it is more likely that the ARDS was due to overwhelming sepsis.

The following are the unique subject identifiers for the three cefepime patients who experienced “cerebral haemorrhage” as a fatal AE.

USUBJID Site ID Sex Age Race Planned Arm 411204016455 016 M 37 W Cefepime

150

411186002007 002 F 62 Cefepime 411189009012 009 M 69 W Cefepime

The M.O. Death Narratives, Patient Profiles, and comments for the preceding three patients follow.

Patient AI411-204-016-455 was a 37 yo White male from the USA with chronic myelogenous leukemia in blast crisis. Recent chemotherapy included: Hydrea (2/91 to (b) (6) ). He developed febrile neutropenia (Temp=39.3 C, ANC=0) on 12/13/93 and received cefepime 2gIVq8h for 9 days ( (b) (6) . Physical exam revealed tachycardia, tachypnea, and epistaxis. Baseline blood and urine cultures were negative. Pertinent Med Hx included: pancytopenia, prolactinoma, psoriasis, epistaxis, visual field loss in the left eye, and elevated liver function tests. Pertinent concomitant therapy included: bacitracin, allopurinol, Parlodel, Tylenol, Demerol, Lasix, electrolyte supplementation, Zofran, and 10 units of PRBCs and 72 units of platelets between (b) (6) Pertinent laboratory results included: at baseline (b) (6) : hemoglobin=6.5, platelets=25,000, creatinine=1.7; on (b) (6) (last recorded): hemoglobin=9.1, platelets=17,000, creatinine=0.7. On Study Day #1, the patient started reinduction chemotherapy (Ara-C and daunorubicin) and was prophylactically placed on fluconazole (100 mg/day) and norfloxacin (800 mg/day). The hospital course was complicated by epistaxis requiring nasal packing and gel foam. By (b) (6) , the patient was afebrile, cefepime was discontinued, and the patient was discharged home. On (b) (6) days post therapy), the patient was readmitted with fever, pancytopenia (WBC=500, platelets=1,000), and coagulopathy. The patient was empirically placed on tobramycin and cefoperazone. The patient was hypotensive (SBP=88) and required dopamine. Physical exam demonstrated multiple sites of bleeding: gingival, retinal, epistaxis. Head CT, obtained due to headache, dizziness, and nausea, showed cerebral hemorrhage. Hemoglobin did not improve despite multiple transfusions. On (b) (6) patient became profoundly hypotensive, developed asystole, and died (14 days post cefepime). No autopsy was performed. The investigator deemed that the patient’s death was caused by cerebral hemorrhage secondary to thrombocytopenia secondary to leukemia and unrelated to cefepime. The following were notable AEs during the course of the study: epistaxis (b) (6) ) of moderate severity that was deemed unrelated to cefepime; cerebral hemorrhage (SAE) ( (b) (6) to death) deemed unrelated to cefepime.

151

M.O. comment: This patient’s death was most likely due to cerebral hemorrhage secondary to prolonged thrombocytopenia secondary to leukemia in blast crisis and concomitant chemotherapy. Though the M.O. cannot completely rule out the possibility that cefepime may have contributed to the prolonged thrombocytopenia and eventual cerebral hemorrhage, other factors mentioned above (blast phase leukemia and chemotherapy) likely played a major role in causing the prolonged thrombocytopenia.

Patient AI411-186-002-007 was a 62 yo female (race unknown, from France) with acute leukemia. Recent chemotherapy (b) (6) included: cytarabine and mitoxantrone. On (b) (6) she developed febrile neutropenia (Temp=39.1 C, ANC=800, [ANC=400 on(b) (6) and received cefepime 2gIVq12h and amikacin IVq12h for 8 days (b) (6) . One out of three baseline blood cultures was positive for E. coli susceptible to study therapy. No additional pertinent Med Hx was noted in the CRF. Pertinent concomitant therapy included: prednisone,

152

zolpidem, paracetamol, codeine, ranitidine, clonazepam, granisetron, metoclopramide, hydrocortisone, phloroglucinol, tiemonium, morphine, nicardipine, PRBCs, platelets, and amphotericin B. No laboratory values were provided in the CRF. On (b) (6) , the subject was (b) afebrile. Follow-up blood cultures through (6) were negative. Thrombocytopenia persisted (b) (b) despite platelet transfusions. On(6) she developed altered mentation and on (6) was diagnosed with severe cerebral hemorrhage and both cefepime and amikacin were discontinued on the same day. The patient died on (b) (6) days post therapy). Autopsy status was not provided in the CRF. The investigator deemed that the patient’s death was due to cerebral hemorrhage secondary to severe thrombocytopenia due to chemotherapy and leukemia and unrelated to study therapy. No SAEs were noted in the CRF.

M.O. comment: This patient’s death was due to cerebral hemorrhage secondary to thrombocytopenia likely from chemotherapy and leukemia. Given that the CRF did not

153

contain baseline and post baseline platelet counts, the M.O. cannot rule out the possibility that study therapy may have contributed to the patient’s thrombocytopenia, though this is a less likely etiology than chemotherapy and acute leukemia.

Patient AI411-189-009-012 was a 69 yo White male from Switzerland with acute myeloid leukemia (AML) (originally diagnosed 3/93). He received chemotherapy (Ara-C and mitoxantrone) from 3/25-3/28. He developed febrile neutropenia on (b) (6) (Temp=38.4 C, ANC=50) and cefepime 2 g IV q8h ( (b) (6) was started on the same day. Baseline cultures were negative for a pathogen. Pertinent Med Hx included: HTN, 1st degree AV block, and myelodysplasia (since 1990). Concomitant therapy included: acyclovir (baseline to (b) (6) , ciprofloxacin 1 gram/day (baseline to (b) (6) and penicillin V 1 million units/day (baseline to(b) (6) , amiodarone, furosemide, prednisone, paracetamol, and terfenadine ( (b) (6) During the hospital course, the fever persisted and the patient developed pulmonary edema, pericardial friction rub, and atrial flutter/fibrillation (b) (6) -death). The patient’s platelet counts were <10,000 from (b) (6) . The patient’s ANC’s were <100 (b) (6) The patient’s AML did not go into remission as evidenced by “massive peripheral blastosis.” The family reportedly elected for supportive care. On (b) (6) the cefepime was discontinued and the patient was started on a ciprofloxacin 1500 mg/day per order of the treating physician. On (b) (6) , the patient developed a cerebral hemorrhage, lapsed into coma, and died 1 hour later. Autopsy was not performed. Serious adverse events included atrial fibrillation and cerebral bleed, and neither was deemed related to study therapy. The investigator deemed that the patient’s death was most likely due to AML, cerebral bleeding, cardiac failure, and rapid atrial fibrillation.

154

M.O. comment: The patient’s death was most likely due to cerebral hemorrhage related to acute leukemia and associated chemotherapy.

The following are the unique subject identifiers for the two cefepime patients who experienced “arrhythmia” as a fatal AE.

USUBJID Site ID Sex Age Race Planned Arm 411204010133 010 M 65 W Cefepime Shands/University 411131001033 of Florida M 43 W Cefepime

The M.O. Death Narratives, Patient Profiles, and comments for the preceding two patients follow.

155

Patient AI411-204-010-133 was a 65 yo White male from the USA with lung cancer (affecting the right middle and upper lobes) and metastatic liver and bone lesions. He received radiation therapy from (b) (6) and chemotherapy with Velban on(b) (6) On (b) (6) , he developed febrile neutropenia (Temp=39.3 C, ANC=295) with cough, dyspnea, tachycardia, chills, and malaise, and was placed on cefepime 2gIVq8h for 5 days (b) (6) . Baseline blood cultures x4, sputum, and urine cultures were negative. CXR demonstrated a right lower lobe lesion. However, initially, the treating physician felt that the patient’s infection source was a skin ulcer on the buttock that grew methicillin-susceptible S. aureus and E. coli. Pertinent Med Hx included: pancytopenia, allergy to Neosporin, and tobacco use. Pertinent concomitant therapy included: vancomycin 2 g/day (b) (6) ), mezlocillin 12 g/day ((b) (6) to death), ceftazidime 3 g/day ((b) (6) to death), erythromycin ( (b) (6) to death), Diflucan ((b) (6) to death), Tylenol, compazine, G-CSF, Proventil, Vanceril, Valium, Thorazine, hydrocortisone, Pepcid, lasix, benadryl, potassium supplementation, Versed, Ativan, Haldol, Dilaudid, Reglan, PRBCs, and platelets. Over the next 2 days, the patient’s fever worsened and CXR demonstrated bilateral pulmonary infiltrates and vancomycin was added. On (b) (6) patient was transferred to the ICU and was placed on mechanical ventilation for respiratory failure. The patient also reportedly developed a tachyarrhythmia of mild severity that resolved in 24 hours. The cefepime was discontinued and replaced with ceftazidime, mezlocillin, and erythromycin. Fluconazole was added on (b) (6) because of Candida sp. in the sputum. According to page 53 of the CRF, “CXR had improved considerably by(b) (6) but pt continued with increased FIO2 requirement (PO2 61 on FIO2 .60 & PEEP 10) and decreased SaO2 with any stress/stimulation. Oncologist at this point felt that cancer was end-stage; family requested discontinuation of life support/therapy on (b) (6) .” The patient died on (b) (6) days post cefepime). No autopsy was performed. The investigator deemed that the patient’s death was caused by the underlying lung cancer and Candida pneumonia, and was not due to cefepime. The following SAEs were noted during the course of the study: hypoxia ( (b) (6) to death) and respiratory failure (b) (6) to death) were deemed unrelated to study therapy.

156

M.O. comment: This patient’s death appeared to ultimately be due to advanced lung cancer. The differential diagnosis for the patient’s lung lesions includes: lung cancer, inflammatory response to chemotherapy, pulmonary edema, Candidal pneumonia, and bacterial pneumonia. The tachyarrhythmia was likely related to the respiratory failure and underlying pulmonary process.

157

Med Hx included: fungal pneumonia 1989 and left vocal cord paralysis. Pertinent concomitant therapy included: Bactrim IV ( (b) (6) ), mycostatin, nystatin, amphotericin B (b) (6) , PRBCs, platelets, hydrocortisone, morphine, CSA (?), Demerol, carafate, aldactone, morphine, electrolyte supplementation, Halcion, compazine, benadryl, Gammagard, Tylenol, lasix, MS contin, Atarax, nifedipine, chloral hydrate, Maalox, albumin, Bumex, Restoril, Dopamine, Flagyl, Ativan, Pentamidine ( (b) (6) only), vancomycin (b) (6) ), and acyclovir (b) (6) Pertinent laboratory values on (b) (6) included: bilirubin=0.9, 1.3, 1.1, 4.4; creatinine=1.0, 1.4, 1.2, 1.5; and BUN=32, 53, 45, 44. Intravenous vancomycin was added on (b) (6) due to the presence of coagulase-negative Staphylococci on blood culture. Amphotericin B (b) (b) (6) was also added on (6) for worsening oral candidiasis. On patient’s fever returned and on (b) (b) (b) (6) he developed altered mentation which cleared after acyclovir was stopped (on (6) ). On (6) (b) Flagyl was added for anaerobic coverage. On (6) the patient developed an erythematous, pustular rash over his entire body. On (b) (6) , a Hickman catheter was removed due to cellulitis at the catheter site and grew C. albicans and Aspergillus fumigatus. Also on (b) (6) , the patient experienced an “irregular heart rhythm with PVC’s” which was deemed of moderate severity and resolved within 3 days. Over the hospital course, the patient continued to deteriorate with rising BUN (45 mg/dL) and bilirubin (4.4 mg/dL), and died on (b) (6) days post therapy). An autopsy was performed and revealed pulmonary aspergillosis. The investigator deemed that the patient’s death was caused by Aspergillus pneumonia and Candidal line sepsis. The following SAE was noted in the CRF: death (b) (6) was deemed unrelated to cefepime.

158

M.O. comment: This patient’s death was most likely due to pulmonary/disseminated Aspergillosis. The arrhythmia was likely related to the underlying pulmonary disease and co- morbid conditions.

The following are the unique subject identifiers for the two cefepime patients who experienced “cardiac arrest” as a fatal AE.

USUBJID Site ID Sex Age Race Planned Arm Shands/University 411131001060 of Florida F 58 W Cefepime 411189010032 010 M 46 W Cefepime

159

The M.O. Death Narratives, Patient Profiles, and comments for the preceding two patients follow.

Patient AI411-131-001-060 was a 58 yo White female from the USA with lymphoblastic lymphoma. Recent chemotherapy (12/28/90-1/10/91) included: vincristine, prednisone, mitoxantrone, and Ara-C. On(b) (6) she developed febrile neutropenia (Temp=101.3 F, ANC<300) and received cefepime 2gIVq8h for 6 days (b) (6) . Baseline blood cultures were negative, and urine culture was positive for Enterococci sp. Pertinent Med Hx included: HTN and vaginal bleeding. Pertinent concomitant therapy included: ceftazidime (b) (6) (b) (6) (b) clotrimazole vaginal suppositories ( ), cefazolin ( (6) only), PRBCs, platelets, allopurinol, Tagamet, prednisone, Mycelex troches, Procardia, Tylenol, benadryl, electrolyte (b) (6) (b) (6) (b) supplementation, Lomotil, Halcion, Flagyl ( ), and Vancomycin On (6) (Day 4), vancomycin IV was added due to persistent fevers and the urine culture positive for enterococci, Flagyl IV was added for presumed intra-abdominal infection, and the patient was transferred to the ICU. The patient’s fevers continued and she developed hypotension on 1/9 (b) with an acute abdomen. Also on(6) (Study Day 6), the patient experienced cardiorespiratory arrest during an abdominal ultrasound and died. No autopsy was performed. The investigator deemed that the patient’s death was caused by an intra-abdominal infection and lymphoblastic lymphoma. The following SAE was noted in the CRF: “death” (b) (6) ) was deemed unrelated to cefepime.

160

161

admission, the patient developed adult respiratory distress syndrome (ARDS) secondary to pulmonary hemorrhage thought secondary to disseminated intravascular coagulation (DIC). Serious adverse events included: ARDS and cardiac arrest, and neither was deemed related to study therapy. The investigator deemed that the patient’s death on (b) (6) was due to underlying leukemia with pulmonary hemorrhage due to DIC and that the MSSA infection was not the primary cause of death.

M.O. comment: The patient’s death was most likely due to ARDS secondary to pulmonary hemorrhage secondary to underlying leukemia and/or MSSA sepsis. The patient died within hours of the first dose of cefepime. The M.O. cannot rule out that the patient’s death was due to study drug failure, though additional factors may have contributed to the patient's death. The patient’s cardiac arrest was likely the sequelae of the ARDS, pulmonary hemorrhage, underlying leukemia, and sepsis.

162

The following are the unique subject identifiers for the five cefepime patients and 1 comparator patient who experienced “renal impairment” as a fatal AE.

USUBJID Site ID Sex Age Race Planned Arm 411186020004 020 F 45 Cefepime 411186020011 020 M 61 Cefepime 411186026009 026 F 39 Cefepime 411186026011 026 F 37 Cefepime 411186004006 004 F 56 Cefepime 411186023001 023 M 36 Ceftazidime

M.O. comment: Of note, all of these patients were from Study 186. In this study the treatment arms were cefepime 2 gIVq12hrs + amikacin versus ceftazidime 2gIVq12hrs + amikacin. Amikacin may also be nephrotoxic.

The M.O. Death Narratives, Patient Profiles, and comments for the preceding six patients follow.

Patient AI411-186-020-004 was a 45 yo female (race unknown, from France) with acute non- lymphocytic leukemia. Recent chemotherapy (4/2-4/11/93) included: cytarabine, lomustine, and doxorubicin. On (b) (6) , she developed febrile neutropenia (Temp=not provided, ANC<400) and received cefepime 2gIVq12h for 14 days ( (b) (6) and amikacin 375mgIVq12h for 16 days ( (b) (6) . Baseline cultures were reportedly negative. No additional pertinent Med Hx was noted in the CRF. Pertinent concomitant therapy included: colistin, neomycin, amphotericin B, nalidixic acid, G-CSF, nicardipine, omeprazole, paracetamol, droperidol, phloroglucinol, PRBCs, and platelets. Vancomycin was added on(b) (6) because of persistent fever. Over the following 2 weeks, the patient manifested signs and symptoms of hepatic failure (icterus, hepatomegaly, tense abdomen, encephalopathy, total bilirubin increased from 12 µmol/L to 318 µmol/L by (b) (6) . The investigator initially suspected veno-occlusive disease. Serum AST, ALT, and alkaline phosphatase reportedly remained within normal limits. On (b) (6) metronidazole and amphotericin B were added for a suspected intra-abdominal infection. On (b) (6) the patient was diagnosed as having life-threatening hepatic failure, severe renal impairment, severe alveolar hemorrhage, was transferred to the ICU, cefepime was discontinued, and imipenem was started. On (b) (6) , amikacin was discontinued due to worsening renal function. On (b) (6) days post cefepime and 2 days post amikacin), the patient died in acute hepatic failure with encephalopathy. No autopsy was performed. The investigator deemed that the patient’s death was due to probable intra-abdominal sepsis (pathogen unknown) and severe aplasia secondary to underlying leukemia and chemotherapy, and unrelated to study therapy. The following SAEs were noted in the CRF: “hepatic failure” ( (b) (6) “renal impairment” (b) (6) and “alveolar hemorrhage” (b) (6) . All were deemed unrelated to study therapy.

163

M.O. comment: This patient’s death was most likely due to acute hepatic failure, renal failure, aplasia, and leukemia. The patient may have also had an underlying intraabdominal infection, though a pathogen was not identified. The M.O. cannot rule out study drug failure or that the study therapy may have contributed to the patient’s hepatic and renal failure. However, it is also likely that the renal and hepatic failure were components and/or sequelae of the patient’s overwhelming multiple system organ failure. The patient was on several drugs that may cause nephrotoxicity, including but not limited to amphotericin B, vancomycin, amikacin, paracetamol, and cefepime.

Patient AI411-186-020-011 was a 61 yo male (race unknown, from France) with acute non- lymphocytic leukemia. Recent chemotherapy (b) (6) -ongoing during study) included: cytarabine. On (b) (6) , he developed febrile neutropenia (Temp=39.0 C, ANC=300) and received cefepime 2gIVq12h for 40 days (b) (6) ) and amikacin IVq12h for 33 days (b) (6)

164

(b) (6) Baseline cultures were negative. CXR was negative on (b) (6) . No additional pertinent Med Hx was noted in the CRF. Pertinent concomitant therapy included: colistin, neomycin, amphotericin B, furosemide, ondansetron, paracetamol, diltiazem, PRBCs, platelets, vitamin K, dexchloropheniramine, fluconazole, and itraconazole. Pertinent laboratory values included: (b) (b) (6) (b) (6) creatinine=115 µmol/L on (6) and 198 µmol/L on . On vancomycin was added because of persistence of fever. On (b) (6) the patient defervesced and remained afebrile for 2 (b) (6) (b) weeks. On , the patient experienced new fevers. Fluconazole was initially added. On (6) the amikacin dose was decreased due to worsening renal function. On (b) (6) , his CXR revealed pulmonary opacities believed to be consistent with pulmonary aspergillosis and amphotericin B was added. On(b) (6) the patient experienced facial paralysis and hemiplegia which were attributed to a suspected Aspergillus brain abscess. On(b) (6) (6 days post cefepime and 13 days post amikacin), the patient died. No autopsy was performed. The investigator deemed that the patient’s death was due to recalcitrant leukemia and disseminated aspergillosis. The following SAEs were noted in the CRF: “facial paralysis (cerebral Aspergillosis)” (b) (6) , “renal failure” ( (b) (6) and “cholestasis” ( (b) (6) . All were deemed unrelated to study therapy.

165

M.O. comment: This patient’s death was most likely due to disseminated Aspergillosis (pulmonary and brain) and chemotherapy-resistant leukemia. The M.O. cannot rule out the possibility that study therapy may have contributed to the patient’s renal failure. The patient was on several drugs that may cause nephrotoxicity, including but not limited to vancomycin, amikacin, paracetamol, and cefepime.

Patient AI411-186-026-009 was a 39 yo female (race unknown, from France) with acute non- lymphocytic leukemia, s/p allogeneic bone marrow transplant on (b) (6) Recent chemotherapy (4/9-4/13/93) included: cytarabine and melphalan; she also received radiotherapy (4/8/93). On (b) (6) she developed febrile neutropenia (Temp=38.8 C, ANC<400) and received cefepime 2gIVq12h for 10 days (b) (6) ) and amikacin 400mgIVq12h for 7 days (b) (6) .

166

Baseline cultures were negative. CXR was negative on (b) (6) . No additional pertinent Med Hx was noted in the CRF. Pertinent concomitant therapy included: cyclosporine, methotrexate, heparin, furosemide, alizapride, ranitidine, alprostadil, morphine, hydrocortisone, dexchloropheniramine, acyclovir, PRBCs, platelets, dopamine, vitamin K, phloroglucinol, chlorazepate, prostaglandin E1, polyvalent immunoglobulin, methylprednisolone, octreotide, ofloxacin, fluconazole, imipenem, and vancomycin. Pertinent laboratory values included: creatinine=62 µmol/L on (b) (6) and 226 µmol/L on (b) (6) total bilirubin=17 µmol/L on (b) (6) and 33 µmol/L on (b) (6) . During the course of the study, acyclovir and vancomycin were initially added for persistent fever though no pathogen was identified on repeat cultures. On (b) (6) , the patient was noted to have worsening hepatic and renal function, and amikacin was discontinued. On (b) (6), cefepime was changed to imipenem due to worsening status and suspected severe sepsis (pathogen not identified). On (b) (6) the patient experienced severe respiratory distress and lapsed into coma. On (b) (6) (2 days post cefepime and 5 days post amikacin), the patient died. Autopsy was reportedly performed; however results were not provided in the CRF. The investigator deemed that the patient’s death was due to multiple organ failure possibly secondary to complications of the bone marrow transplant and unrelated to study therapy. The following SAEs were noted in the CRF: “renal impairment” (b) (6) ) and “hepatic impairment” (b) (6) (b) (6) were deemed unrelated to study therapy.

167

M.O. comment: This patient’s death was due to multiple organ failure. The differential diagnosis includes: failed bone marrow transplant/veno-occlusive disease, acute leukemia, sepsis due to an unknown pathogen (cannot rule out study therapy failure), and drug toxicity (study therapy and/or concomitant medications). The M.O. cannot rule out the possibility that study therapy may have contributed to the patient’s hepatic and renal failure. The patient was on several drugs that may cause nephrotoxicity, including but not limited to cyclosporine, amikacin, acyclovir, and cefepime.

Patient AI411-186-026-011 was a 37 yo female (race unknown, from France) with acute lymphoblastic leukemia, s/p bone marrow transplant on (b) (6) (Day 3 of study therapy). Recent chemotherapy (4/30-5/4/93) included: cytarabine and melphalan; she also received radiotherapy (b) (4/26-4/29/93). On (6) , she developed febrile neutropenia (Temp=39.8 C, ANC<200) and received cefepime 2gIVq12h for 10 days (b) (6) and amikacin 500mgIVq12h for 4 days (b) (6)

168

(b) (6) . Baseline blood cultures were negative. CXR was negative. No additional pertinent Med Hx was noted in the CRF. Pertinent concomitant therapy included: heparin, cyclosporine, alizapride, ranitidine, furosemide, ondansetron, octreotide, clobutinol, loperamide, fluoxetine, tropatepine, platelets, PRBCs, morphine, pethidine, phloroglucinol, chlorazepate, methotrexate, G-CSF, dopamine, polyvalent immunoglobulin, fluconazole, and amphotericin B. Pertinent (b) (b) (6) laboratory values included: serum creatinine was 79 µmol/L on (6) and 315 µmol/L on On (b) (6)the patient underwent bone marrow transplant and was placed on methotrexate and cyclosporine as prophylaxis against graft-versus-host disease. IV acyclovir was also added. (b) Fever persisted despite negative cultures. On (6) amikacin was discontinued due to worsening (b) renal function and ciprofloxacin was added. On (6) , a single blood culture grew a Streptococcus sp. (per CRF, susceptibilities were not performed because the investigator considered this a “contaminant”) and vancomycin was added. On (b) (6) the patient developed signs of shock and multiple organ failure, and was transferred to the ICU. On (b) (6) , cefepime was changed to imipenem and on (b) (6) amphotericin B was added. The patient died on (b) (6) days post cefepime and 20 days post amikacin). No autopsy was performed. The investigator deemed that the patient’s death was due to sepsis (pathogen unknown), multiple organ failure, and renal failure due to methotrexate and cyclosporine administration, and unrelated to study therapy. The following SAE was noted in the CRF: “renal impairment” (b) (6) was deemed unrelated to study therapy.

169

M.O. comment: This patient’s death was due to multiple organ failure. The differential diagnosis includes: severe sepsis/septic shock due to an unknown pathogen (possibly a viridans group Streptococci, cannot rule out study therapy failure), failed bone marrow transplant with graft versus host disease, acute leukemia, and drug toxicity (study therapy and/or concomitant medications). The M.O. cannot rule out the possibility that study therapy may have contributed to the patient’s renal failure. However, the patient was on several drugs that may cause nephrotoxicity, including but not limited to cyclosporine, amikacin, acyclovir, and cefepime. The M.O. also cannot rule out the possibility that this patient may have developed serotonin syndrome because this patient was on a number of agents linked to this syndrome (fluoxetine, pethidine, morphine, alizapride, and tropatepine).

170

Patient AI411-186-004-006 was a 56 yo female (race unknown, from France) with acute non Hodgkin’s lymphoma who received an autologous bone marrow transplant on (b) (6) Recent chemotherapy ( (b) (6) included: cytarabine and melphalan; she also received radiotherapy ( (b) (6) On(b) (6) she developed febrile neutropenia (Temp=40.0 C, ANC<20) and received cefepime 2gIVq12h for 7 days (b) (6) and amikacin 650 mgIVq12h for 5 days (b) (6) Baseline urine culture grew E. coli and blood cultures were negative. CXR was negative on (b) (6) No additional pertinent Med Hx was noted in the CRF. Pertinent concomitant therapy included: antimicrobials as discussed below, pefloxacin, fluconazole, metoclopramide, methylprednisolone, ondansetron, heparin, paracetamol, furosemide, insulin, dopamine, PRBCs, platelets, G-CSF, bromhexidine, and ganciclovir. The patient’s condition progressively worsened over the course of the hospitalization. On (b) (6) she remained febrile, and began developing respiratory and renal dysfunction. Vancomycin and acyclovir were added and the patient was transferred to the ICU for mechanical ventilation. On (b) (6) , the patient underwent bronchoscopy. BAL culture grew respiratory syncytial virus (RSV). Amikacin was stopped on (b) (6) and cefepime was changed to piperacillin/tazobactam on (b) (6) . Repeat urine culture on(b) (6) was negative. The patient was placed on aerosolized ribavirin for RSV; however, CXRs demonstrated progressively worsening interstitial pneumonia. The patient died on (b) (6) (29 days post cefepime and 31 days post amikacin). No autopsy was performed. The investigator deemed that the patient’s death was due to respiratory failure caused by RSV pneumonia and pancytopenia following chemotherapy. The following SAEs were noted in the CRF: “RSV pneumonia” (b) (6) ) and “acute renal insufficiency” requiring hemodialysis ((b) (6) were deemed unrelated to study therapy.

171

M.O. comment: This patient’s death was most likely due to RSV pneumonia. The M.O. cannot rule out the possibility that study therapy contributed to the patient’s renal failure. The patient was on several drugs that may cause nephrotoxicity, including but not limited to vancomycin, paracetamol, amikacin, acyclovir, and cefepime.

Patient AI411-186-023-001 was a 36 yo male (race unknown, from France) with Stage IV non Hodgkin’s lymphoma, s/p autologous bone marrow transplant on (b) (6) Recent chemotherapy (3/11-3/16/93) included: carmustine, etoposide, cytarabine, and melphalan. On(b) (6) he developed febrile neutropenia (Temp=38.8 C, ANC<100) and received ceftazidime 2gIVq8h for 2 days (b) (6) and amikacin 600mgIVq12h for 11 days (b) (6) Baseline blood cultures were negative. CXR showed a right-sided pneumonia. No additional pertinent Med Hx was noted in the CRF. Pertinent concomitant therapy included: antimicrobial agents as noted below, oral fluconazole, gentamicin, and neomycin for gut decontamination; loperamide, dopamine,

172

dobutamine, furosemide, benzodiazepines, Hordenol (for diarrhea), methylprednisolone, ranitidine, and bumetanide. Pertinent laboratory values included: creatinine=124 µmol/L on (b) (6) (b) (b) (6) (b) (6) (b) (6) and 288 µmol/L on (6) , total bilirubin=16 µmol/L on and 127 µmol/L on On , the patient developed septic shock with cardiac failure and ARDS and was transferred to the ICU. Dopamine and dobutamine were initiated. CXR revealed bilateral pneumonia and cardiac echo revealed “signs of fluid overload”. Ceftazidime was stopped and imipenem and vancomycin were started on the same day. The dose of amikacin was increased to 750mgIVq12h. On (b) (6) the patient developed renal failure requiring hemodialysis. On (b) (6) the patient underwent bronchoscopy and BAL culture grew MRSA. The patient developed multiple organ failure and died on (b) (6) (9 days post ceftazidime and on therapy Day 11 of amikacin). No autopsy was performed. The investigator deemed that the patient’s death was due to MRSA pneumonia and septic shock with resultant renal failure and lactic acidosis. The following SAEs were noted in the CRF: cardiac failure (b) (6) ), renal insufficiency (b) (6) respiratory insufficiency (b) (6) (b) (b) (6) , and lactic acidosis ( (6) ). All were deemed unrelated to study therapy.

173

M.O. comment: This patient’s death was most likely due to MRSA pneumonia and septic shock. Renal failure and lactic acidosis were likely due to the underlying septic process; however, the M.O. cannot rule out that study therapy may have contributed to the renal and hepatic failure. The patient was on several drugs that may cause nephrotoxicity, including but not limited to vancomycin, amikacin, imipenem, and ceftazidime.

174

Analyses based on concomitant medical history and 30-day mortality (collaborative work with Ana Szarfman, M.D., Ph.D.)

The following two tables summarize the analyses of the presence or absence of acute leukemia, solid tumors, and lymphoma or multiple myeloma as co-morbid conditions and 30-day mortality among patients enrolled in the nine febrile neutropenia studies.

175

Table 9. Summary of Frequency of Co-morbid Conditions by Treatment and Study for the 9 Febrile Neutropenia (using last randomized treatment)

Study With Acute Leukemia With Solid Tumors With Lymphoma/MM Total No. of Patients/Study

Cefepime Comparator Cefepime Comparator Cefepime Comparator Cefepime Comparator Total % % % % % n % (n/N) n (n/N) n (n/N) n (n/N) n (n/N) n (n/N) N N N

ai411118 Cefepime vs Piperacillin + Gentamicin 1:1 20 33.90 16 28.07 17 28.81 23 40.35 21 35.59 17 29.82 59 57 116

ai411131 Cefepime vs Ceftazidime 1:1 60 57.69 46 51.11 27 25.96 25 27.78 13 12.50 13 14.44 104 90 194

ai411137 Cefepime vs Mezlocillin + Gentamicin 1:1 0 0.00 2 5.56 4 11.43 4 11.11 26 74.29 25 69.44 35 36 71

ai411143 Uncontrolled 30 35.71 0 29 34.52 0 19 22.62 0 84 0 84

ai411158 Uncontrolled 14 46.67 0 0 0.00 0 6 20.00 0 30 0 30

ai411186 Cefepime + Amikacin vs Ceftazidime + Amikacin 2:1 136 56.20 61 54.95 0 0.00 0 0.00 87 35.95 41 36.94 242 111 353

ai411189 Cefepime vs Ceftazidime 1:1 68 47.22 66 48.18 24 16.67 17 12.41 45 31.25 43 31.39 144 137 281

ai411198 Cefepime + Vancomycin vs Ceftazidime + Vancomycin 1:1 28 51.85 26 45.61 2 3.70 1 1.75 15 27.78 20 35.09 54 57 111

ai411204 Cefepime vs Ceftazidime 1:1 31 22.46 29 21.01 67 48.55 67 48.55 31 22.46 31 22.46 138 138 276

Total only Comparative Studies 343 44.20 246 39.30 141 18.17 137 21.88 238 30.67 190 30.35 776 626 1402 Total All Studies 387 43.48 246 39.30 170 19.10 137 21.88 263 29.55 190 30.35 890 626 1516 *Additional patients beyond those captured in the selected co-morbid condition categories were enrolled in these studies. Therefore, the patient totals per co-morbid condition will not necessarily sum up to the treatment arm columns under "Total No. of Patients/Study".

M.O. comment: Please refer to the following sections of this review for additional analyses based on each of the co-morbid conditions noted in this table.

176

Table 10. Summary of Frequency of Co-morbid Conditions and 30-Day Mortality by Treatment and Study for the 9 Febrile Neutropenia (using last randomized treatment)

Study Deaths With Acute Leukemia With Solid Tumors With Lymphoma/MM Total No. of Deaths/Study

Cefepime Comparator Cefepime Comparator Cefepime Comparator Cefepime Comparator Total % % % % % n % (n/N) n (n/N) n (n/N) n (n/N) n (n/N) n (n/N) N N N ai411118 Cefepime vs Piperacillin + Gentamicin 1:1 2 40.00 2 50.00 2 40.00 0 0.00 1 20.00 2 50.00 5 4 9 ai411131 Cefepime vs Ceftazidime 1:1 5 55.56 4 66.67 3 33.33 1 16.67 1 11.11 0 0.00 9 6 15 ai411137 Cefepime vs Mezlocillin + Gentamicin 1:1 0 0.00 0 0.00 0 0 0.00 0 1 100.00 0 1 1 ai411143 Uncontrolled 2 25.00 0 0.00 4 50.00 0 0.00 0 0.00 0 0.00 8 0 8 ai411158 Uncontrolled 1 25.00 0 0.00 0 0.00 0 0.00 0 0.00 0 0.00 4 0 4 ai411186 Cefepime + Am kacin vs Ceftazidime + Amikacin 2:1 10 76.92 1 25.00 0 0.00 0 0.00 3 23.08 3 75.00 13 4 17 ai411189 Cefepime vs Ceftazidime 1:1 3 37.50 3 30.00 1 12.50 0 0.00 3 37.50 5 50.00 8 10 18 ai411198 Cefepime + Vancomycin vs Ceftazidime + Vancomycin 1:1 4 80.00 0 0.00 0 0.00 0 0.00 1 20.00 2 66.67 5 3 8 ai411204 Cefepime vs Ceftazidime 1:1 5 23.81 4 30.77 8 38.10 4 30.77 5 23.81 3 23.08 21 13 34

Total only Comparative Studies 29 47.54 14 34.15 14 22.95 5 12.20 14 22.95 16 39.02 61 41 102

Total All Studies 32 43.84 14 34.15 18 24.66 5 12.20 14 19.18 16 39.02 73 41 114

*Additional patients beyond those captured in the selected co-morbid condition categories were enrolled in these studies. Ther efore,the patient totals per co-morbid condition will not necessarily sum up to the treatment arm columns under "Total No. of Patients/Study".

M.O. comment: Please refer to the following sections of this review for additional analyses based on each of the co-morbid conditions noted in this table.

177

The following two tables provide analyses of the presence or absence of acute leukemia as a co-morbid condition and 30-day mortality among patients enrolled in the nine febrile neutropenia studies.

Table 11. Frequency of Acute Leukemia by Treatment and Study for the 9 Febrile Neutropenia Studies (using last randomized treatment) Study With Acute Leukemia Total No. of Patients/Study Cefepime Comparator Cefepime Comparator Total n % (n/N) n % (n/N) N N N ai411118 Cefepime vs Piperacillin + Gentamicin 1:1 20 33.90 16 28.07 59 57 116 ai411131 Cefepime vs Ceftazidime 1:1 60 57.69 46 51.11 104 90 194 ai411137 Cefepime vs Mezlocillin + Gentamicin 1:1 0 0.00 2 5.56 35 36 71 ai411143 Uncontrolled 30 35.71 0 0.00 84 0 84 ai411158 Uncontrolled 14 46.67 0 0.00 30 0 30 ai411186 Cefepime + Amikacin vs Ceftazidime + Amikacin 2:1 136 56.20 61 54.95 242 111 353 ai411189 Cefepime vs Ceftazidime 1:1 68 47.22 66 48.18 144 137 281 ai411198 Cefepime + Vancomycin vs Ceftazidime + Vancomycin 1:1 28 51.85 26 45.61 54 57 111 ai411204 Cefepime vs Ceftazidime 1:1 31 22.46 29 21.01 138 138 276 Total only Comparative Studies 343 44.20 246 39.30 776 626 1402 Total All Studies 387 43.48 246 39.30 890 626 1516

M.O. comment: Regarding the comparative studies, and based on the current analysis, there were slightly more acute leukemia patients treated with cefepime versus comparator. In particular, this was noted for Studies 118, 131, 186, 198, and 204.

178

Table 12. Frequency of Acute Leukemia and 30-Day Mortality by Treatment and Study in the 9 Febrile Neutropenia Studies (using last randomized treatment) Study Deaths With Acute Leukemia Total No. of Deaths/Study Cefepime Comparator Cefepime Comparator Total n % (n/N) n % (n/N) N N N ai411118 Cefepime vs Piperacillin + Gentamicin 1:1 2 40.00 2 50.00 5 4 9 ai411131 Cefepime vs Ceftazidime 1:1 5 55.56 4 66.67 9 6 15 ai411137 Cefepime vs Mezlocillin + Gentamicin 1:1 0 0.00 0 0.00 0 1 1 ai411143 Uncontrolled 2 25.00 0 0.00 8 0 8 ai411158 Uncontrolled 1 25.00 0 0.00 4 0 4 ai411186 Cefepime + Amikacin vs Ceftazidime + Amikacin 2:1 10 76.92 1 25.00 13 4 17 ai411189 Cefepime vs Ceftazidime 1:1 3 37.50 3 30.00 8 10 18 ai411198 Cefepime + Vancomycin vs Ceftazidime + Vancomycin 1:1 4 80.00 0 0.00 5 3 8 ai411204 Cefepime vs Ceftazidime 1:1 5 23.81 4 30.77 21 13 34 Total only Comparative Studies 29 47.54 14 34.15 61 41 102 Total All Studies 32 43.84 14 34.15 73 41 114

M.O. comment: Regarding the comparative studies, a higher proportion of cefepime versus comparator patients with acute leukemia died.

179

The following two tables provide analyses of the presence or absence of a solid tumor as a co-morbid condition and 30-day mortality among patients enrolled in the nine febrile neutropenia studies.

Table 13. Frequency of Solid Tumors by Treatment and Study for the 9 Febrile Neutropenia (using last randomized treatment) Study With Solid Tumors Total No. of Patients/Study Cefepime Comparator Cefepime Comparator Total n % (n/N) n % (n/N) N N N ai411118 Cefepime vs Piperacillin + Gentamicin 1:1 17 28.81 23 40.35 59 57 116 ai411131 Cefepime vs Ceftazidime 1:1 27 25.96 25 27.78 104 90 194 ai411137 Cefepime vs Mezlocillin + Gentamicin 1:1 4 11.43 4 11.11 35 36 71 ai411143 Uncontrolled 29 34.52 0 0.00 84 0 84 ai411158 Uncontrolled 0 0.00 0 0.00 30 0 30 ai411186 Cefepime + Amikacin vs Ceftazidime + Amikacin 2:1 0 0.00 0 0.00 242 111 353 ai411189 Cefepime vs Ceftazidime 1:1 24 16.67 17 12.41 144 137 281 ai411198 Cefepime + Vancomycin vs Ceftazidime + Vancomycin 1:1 2 3.70 1 1.75 54 57 111 ai411204 Cefepime vs Ceftazidime 1:1 67 48.55 67 48.55 138 138 276 Total only Comparative Studies 141 18.17 137 21.88 776 626 1402 Total All Studies 170 19.10 137 21.88 890 626 1516

M.O. comment: Regarding the comparative studies, overall, a higher proportion of solid tumor patients were treated with comparator versus cefepime.

180

Table 14. Frequency of Solid Tumors and 30-Day Mortality by Treatment and Study in the 9 Febrile Neutropenia Studies (using last randomized treatment) Study Deaths With Solid Tumors Total No. of Deaths/Study Cefepime Comparator Cefepime Comparator Total n % (n/N) n % (n/N) N N N ai411118 Cefepime vs Piperacillin + Gentamicin 1:1 2 40.00 0 0.00 5 4 9 ai411131 Cefepime vs Ceftazidime 1:1 3 33.33 1 16.67 9 6 15 ai411137 Cefepime vs Mezlocillin + Gentamicin 1:1 0 0 0.00 0 1 1 ai411143 Uncontrolled 4 50.00 0 0.00 8 0 8 ai411158 Uncontrolled 0 0.00 0 0.00 4 0 4 ai411186 Cefepime + Amikacin vs Ceftazidime + Amikacin 2:1 0 0.00 0 0.00 13 4 17 ai411189 Cefepime vs Ceftazidime 1:1 1 12.50 0 0.00 8 10 18 ai411198 Cefepime + Vancomycin vs Ceftazidime + Vancomycin 1:1 0 0.00 0 0.00 5 3 8 ai411204 Cefepime vs Ceftazidime 1:1 8 38.10 4 30.77 21 13 34 Total only Comparative Studies 14 22.95 5 12.20 61 41 102 Total All Studies 18 24.66 5 12.20 73 41 114

M.O. comment: Regarding the comparative studies, a higher proportion of cefepime versus comparator patients with solid tumors died. However, the overall number of patients who died with the diagnosis of solid tumor was relatively low.

The following two tables provide analyses of the presence or absence of a lymphoma or multiple myeloma as co-morbid conditions and 30-day mortality among patients enrolled in the nine febrile neutropenia studies.

181

Table 15. Frequency of Lymphoma or Multiple Myeloma by Treatment and Study for the 9 Febrile Neutropenia (using last randomized treatment)

Study With Lymphoma/MM Total No. of Patients/Study Cefepime Comparator Cefepime Comparator Total n % (n/N) n % (n/N) N N N ai411118 Cefepime vs Piperacillin + Gentamicin 1:1 21 35.59 17 29.82 59 57 116 ai411131 Cefepime vs Ceftazidime 1:1 13 12.50 13 14.44 104 90 194 ai411137 Cefepime vs Mezlocillin + Gentamicin 1:1 26 74.29 25 69.44 35 36 71 ai411143 Uncontrolled 19 22.62 0 0.00 84 0 84 ai411158 Uncontrolled 6 20.00 0 0.00 30 0 30 ai411186 Cefepime + Amikacin vs Ceftazidime + Amikacin 2:1 87 35.95 41 36.94 242 111 353 ai411189 Cefepime vs Ceftazidime 1:1 45 31.25 43 31.39 144 137 281 ai411198 Cefepime + Vancomycin vs Ceftazidime + Vancomycin 1:1 15 27.78 20 35.09 54 57 111 ai411204 Cefepime vs Ceftazidime 1:1 31 22.46 31 22.46 138 138 276 Total only Comparative Studies 238 30.67 190 30.35 776 626 1402 Total All Studies 263 29.55 190 30.35 890 626 1516

M.O. comment: Regarding the comparative studies, overall, there were similar proportions of multiple myeloma/lymphoma patients treated with cefepime versus comparator.

182

Table 16. Frequency of Lymphoma or Multiple Myeloma and 30-Day Mortality by Treatment and Study in the 9 Febrile Neutropenia

Studies (using last randomized treatment)

Study Deaths With Lymphoma/MM Total No. of Deaths/Study Cefepime Comparator Cefepime Comparator Total n % (n/N) n % (n/N) N N N ai411118 Cefepime vs Piperacillin + Gentamicin 1:1 1 20.00 2 50.00 5 4 9 ai411131 Cefepime vs Ceftazidime 1:1 1 11.11 0 0.00 9 6 15 ai411137 Cefepime vs Mezlocillin + Gentamicin 1:1 0 1 100.00 0 1 1 ai411143 Uncontrolled 0 0.00 0 0.00 8 0 8 ai411158 Uncontrolled 0 0.00 0 0.00 4 0 4 ai411186 Cefepime + Amikacin vs Ceftazidime + Amikacin 2:1 3 23.08 3 75.00 13 4 17 ai411189 Cefepime vs Ceftazidime 1:1 3 37.50 5 50.00 8 10 18 ai411198 Cefepime + Vancomycin vs Ceftazidime + Vancomycin 1:1 1 20.00 2 66.67 5 3 8 ai411204 Cefepime vs Ceftazidime 1:1 5 23.81 3 23.08 21 13 34 Total only Comparative Studies 14 22.95 16 39.02 61 41 102 Total All Studies 14 19.18 16 39.02 73 41 114

M.O. comment: Regarding the comparative studies, overall, a higher proportion of comparator patients (39%) with lymphoma/multiple myeloma died as compared with cefepime-treated patients (23%). Of note, overall numbers of patients with lymphoma/multiple myeloma were relatively low.

The following figure provides a meta-analysis of the febrile neutropenia study data as run by Ana Szarfman, M.D., Ph.D., based on Multivariate Bayesian Logistic Regression (MBLR) using a program developed by William DuMochel, Ph.D., from Lincoln Technologies.

183

Figure 8. MBLR Analysis of Variables Based on Association with 30-day Mortality and Study Treatment for the 9 Febrile Neutropenia Studies. (Note: Indication = Study)

184

M.O. comment: Of note, point estimates to the right of 1.0 indicate increased association of cefepime with 30-day mortality, and point estimates to the left of 1.0 indicate increased associated of comparator agents with 30-day mortality. Based on this analysis, none of the proposed risk factors, in conjunction with study treatment (cefepime versus comparators), were statistically significantly associated with 30-day mortality. Additionally, Dr. Szarfman and the M.O. performed a MBLR analysis of solely the 7 comparative studies and the adjusted odds ratio for death for cefepime vs. comparator patients was 1.36 (95% CI: 0.28, 3.60). Please refer to the review by Ana Szarfman, M.D., Ph.D., for additional analysis and information related to the construction of the logistic regression model.

Overall Conclusions of Ana Szarfman, M.D., Ph.D.: Dr. Szarfman concluded that based on her Multivariate Bayesian Logistic Regression (MBLR) data mining analysis of the 9 BMS-sponsored febrile neutropenia studies (7 comparative and 2 non-comparative), cefepime as studied (empiric therapy for febrile neutropenia) was not associated with a significantly higher mortality rate versus the comparator group, i.e., she did not confirm the findings of Yahav et al.1 Dr. Szarfman identified very small and non-significant effects of cefepime treatment on death in several subgroups. These subgroups included the oldest age group analyzed, whites, studies ai411204, ai411131, ai411198, ai411186, patients characterized by the applicant as having received medications for a bone marrow transplant, patients with diabetes mellitus, and patient with baseline serum creatinine above 2.5 mg/dL and with neutrophil counts below 500 U/uL.

185

[SEE THE SECOND ADDENDUM ATTACHED TO THIS REVIEW.]

4.112 HIGHLIGHTS FROM THE ANALYSIS PERFORMED BY YU-TE WU, PH.D. (STATISTICAL REVIEWER, QUANTITATIVE SAFETY AND PHARMACOEPIDEMIOLOGY GROUP)

The Quantitative Safety and Pharmacoepidemiology group was consulted to perform a new meta-analysis based not only on trial level mortality data from Yahav et al.1, but also to include mortality data from: (1) BMS studies originally submitted to the U.S. FDA that may not have been submitted for publication and were therefore omitted from the Yahav analysis, as well as, (2) unpublished cefepime studies that were never submitted to the U.S. FDA. In addition, the meta-analysis team reviewed some patient level data across multiple studies as per an agreed upon statistical analysis plan. This analysis was mainly performed by Yu-te Wu, Ph.D. Specifically, she performed the following analyses. • Dr. Wu repeated the Yahav et al.1 meta-analysis to ensure that the authors’ results were replicable.

Please refer to Dr. Wu’s review for additional detail and analyses beyond what is noted in the current M.O. review.

In a trial level analysis of 88 studies that included 9,467 cefepime-treated patients and 8,288 comparator-treated patients, the overall mortality rates were 6.21% (588/9467) for cefepime treated patients and 6.00% (497/8288) for comparator-treated patients [Risk Difference = 5.38 (95% C.I.: -1.53,12.28), see Figure 9]. The following Table 17 delineates the numbers of patients and studies analyzed by treatment indication.

Table 17. Trial Level Analysis of Patient Numbers and Studies by Treatment Indication

Indications Number of Number of subjects Studies Cefepime (%) Comparator (%) Febrile Neutropenia 24 186 2791(29.48%) 2658(32.07%) Intra-abdominal Infection 7 628(6.63%) 470(5.67%) Pneumonia 26 2228(23.53%) 1821(21.97%) Urinary Tract Infection 7 763(8.06%) 490(5.91%) Skin Structure Infection 2 335(3 54%) 165(1 99%) [THE COMPLETE TABLE 17 IS ATTACHED TO THIS REVIEW AS AN ADDENDUM]

M.O. comment: The clinical conditions (listed as “indications” in the current and following tables) with the most studies represented in Dr. Wu’s trial level analysis were: Pneumonia, Febrile Neutropenia, and “Other”. The “Other” category included, but were not limited to studies for the treatment of bacterial meningitis, bacterial endocarditis, bloodstream infections, and mixed infections (studies of patients with a variety of infections).

The following Table 18 provides the numbers of patients per study that used specific comparator agents.

Table 18. Trial Level Analysis of Patient Numbers per Study that Utilized Specific Comparator Agents Type of Comparator Drug Number of Number of subjects Studies Cefepime (%) Comparator (%) Ceftazidime 47 5430(57.36%) 4580(55.26%) Piperacillin-tazobactam 6 965(10.19%) 976(11.78%) Imipenem-meropenem 6 782(8.26%) 749(9.04%) Ceftriaxone-cefotaxime 17 1270(13.42%) 1140(13.75%) Other 11 818(8.64%) 641(7.73%) Mixed treatment 1 202(2.13%) 202(2.44%) Total 88 9467 8288

M.O. comment: The comparator agents with the most studies represented in Dr. Wu’s trial level analysis were: ceftazidime, ceftriaxone-cefotaxime, and “other”. Examples of “other” regimens included: mezlocillin, mezlocillin+gentamicin, cefuroxime, sulbactam+cefoperazone, amikacin, and clindamycin+gentamicin.

The following Figure 9 provides Dr. Wu’s analysis of 30-day mortality stratified by study drug (cefepime versus comparator) and treatment indication.

187

Figure 9. 30-Day Mortality Stratified by Study Drug (Cefepime versus Comparator) and Treatment Indication.

M.O. comment: Overall, there was not a statistically significant difference in 30-day mortality for patients treated with cefepime versus comparator agents. Aside from the skin structure infection indication [which contained a relatively small number of patients and deaths (6 deaths for cefepime versus 0 for comparators)], the analysis did not demonstrate a statistically significant difference in 30-day mortality for cefepime versus comparator agents for any of the other treatment indications analyzed.

188

Figure 10. 30-Day Mortality Stratified by Comparator Agents versus Cefepime.

M.O. comment: The analysis did not demonstrate a statistically significant difference in 30- day mortality for cefepime versus any of the comparator agents, except for ceftazidime. It is unclear why ceftazidime administration would be associated with less mortality as compared with any of the other comparator drugs in the analysis.

In addition, Dr. Wu performed a patient level analysis with 35 BMS-sponsored studies for which such data was available. The analysis population included 5058 cefepime patients and 3976 comparator patients. Thirty-day mortality was 5.63% (285/5058) for cefepime and 5.68% (226/3976) for comparator agents.

The patient level data were derived from the following studies.

Table 19. Patient Level Analysis of Patient Numbers and Studies by Treatment Indication

Indications Number of Number of subjects Studies Cefepime (%) Comparator (%) Febrile Neutropenia 7 776(15.34%) 626(15.74%) Intra-abdominal Infection 5 585(11.57%) 421(10.59%) Pneumonia 4 609(12.04%) 306(7.70%) Urinary Tract Infection 4 426(8.42%) 242(6.09%) Skin Structure Infection 2 335(6.62%) 165(4.15%) Other 13 2327(46.01%) 2216(55.73%) Total 35 5058 3976 M.O. comment: The majority of studies were derived from the “Other” treatment indication, febrile neutropenia, and intra-abdominal infections.

189

Table 20. Patient Level Analysis of Patient Numbers and Studies Based on Comparator Agent

Type of Comparator Drug Number of Number of subjects Studies Cefepime (%) Comparator (%) Ceftazidime 21 3617(71.51%) 2839(71.40%) Piperacillin-tazobactam 1 59(1.17%) 57(1.43%) Imipenem-meropenem 1 164(3.24%) 159(4.00%) Ceftriaxone-cefotaxime 9 794(15.70%) 654(16.45%) Other 4 424(8.38%) 267(6.72%) Total 36 5058 3976

M.O. comment: The majority of the studies contained ceftazidime as a comparator agent.

Dr. Wu noted that there were no notable differences in mortality between patients taking cefepime versus comparator drugs for age, gender, race, baseline characteristics (study location, any pathogen recovered at baseline, all pathogens isolated at baseline susceptible to study therapy, fungal pathogen recovered at baseline, baseline infection (mono- or polymicrobial), renal insufficiency or failure, active cancer or malignancy, and history of bone marrow transplant), and treatment duration.

The following figures provide additional details.

Figure 11. Patient Level Analysis of Age versus 30-Day Mortality Stratified on Treatment Group (Cefepime vs. Comparator)

190

M.O. comment: The analysis by age did not demonstrate a statistically significant difference in 30-day mortality between cefepime and comparator regimens.

Figure 12. Patient Level Analysis of Gender versus 30-Day Mortality Stratified on Treatment Group (Cefepime vs. Comparator)

191

M.O. comment: The analysis by gender did not demonstrate a statistically significant difference in 30-day mortality between cefepime and comparator regimens.

Figure 13. Patient Level Analysis of Race versus 30-Day Mortality Stratified on Treatment Group (Cefepime vs. Comparator)

M.O. comment: The analysis by race did not demonstrate a statistically significant difference in 30-day mortality between cefepime and comparator regimens.

192

Figure 14. Patient Level Analysis of Geographic Region (U.S vs. non-U.S) versus 30-Day Mortality Stratified on Treatment Group (Cefepime vs. Comparator)

M.O. comment: The analysis by geographic distribution (U.S. versus non-U.S. based studies) did not demonstrate a statistically significant difference in 30-day mortality between cefepime and comparator regimens.

Figure 15. Patient Level Analysis of Whether or Not a Pathogen was Recovered at Baseline versus 30-Day Mortality Stratified on Treatment Group (Cefepime vs. Comparator)

193

M.O. comment: The analysis by recovery of pathogen at baseline did not demonstrate a statistically significant difference in 30-day mortality between cefepime and comparator regimens.

Figure 16. Patient Level Analysis of Baseline Pathogen Susceptibility to Study Drug versus 30 Day Mortality Stratified on Treatment Group (Cefepime vs. Comparator)

M.O. comment: Overall, the analysis of the susceptibility of the baseline pathogen(s) to study drug did not demonstrate a statistically significant difference in 30-day mortality between

194

cefepime and comparator regimens. However, cefepime-treated patients with a resistant pathogen at baseline were more likely to die within 30 days of study therapy than comparator- treated patients. This led to an information request to BMS that included: (1) a request for additional case report forms for 19/40 patients who had a resistant pathogen at baseline and died within 30 days post therapy (these 19 patients’ CRFs had not yet been provided by BMS); and (2) a dataset and summary table on patients from comparative cefepime studies with baseline pathogens resistant to study therapy. In addition, the Agency held a teleconference with BMS on 11/21/08 to further understand why more cefepime patients with a baseline resistant pathogen appeared to died as compared to comparator patients. As discussed during the teleconference and confirmed in a written submission from dated 12/12/08, they noted that “dirty/missing data” may have led to the discrepancies noted above.

The following on page 2 of their response.

In addition, BMS noted the following on further inspection of their datasets.

195

Based on the revised dataset, Dr. Wu created the following summary tables on the 423 patients who had a pathogen resistant to study therapy at baseline. If study therapy included a combination regimen, patients were included if the pathogen(s) was (were) resistant to either agent in the combination.

196

Table 21. Mortality Stratified by Indications, Demographics, and Treatment Arm for 423 Patients with a Resistant Pathogen at Baseline

Alive Death Total Cefepime Controls Total Cefepime Controls Indication Febrile neutropenia 86(22.45) 46(21.50) 40(23.67) 7(17.50) 4 (12.90) 3(33.33) Intra-abdominal 81(21.15) 37(17.29) 44(26.04) 0 0 0 LRTI 7(1.83) 5(2.34) 2(1.18) 2(5.00) 2 (6.45) 0 UTI 15(3.92) 8(3.74) 7(4.14) 2(5.00) 2 (6.45) 0 SSTI 21(5.48) 12(5.61) 9(5.33) 1(2.50) 1 (3.23) 0 Mixed infect 90(23.50) 53(24.77) 37(21.89) 24(60.00) 19(61.29) 5(55.56) Other 83(21.67) 53(24.77) 30(17.75) 4(10.00) 3(9.68) 1(11.11) Total 383 214 169 40 31 9

Sex Female 174(45.43) 113(52.80) 61(36.09) 13(32.50) 10(32.26) 3(33.33) Male 209(54.57) 101(47.20) 108(63.91) 27(67.50) 21(67.74) 6(66.67) Total 383 214 169 40 31 9

Race Black 55(14.36) 33(15.42) 22(13.02) 3(7.50) 2(6.45) 1(11.11) Hispanic 69(18.02) 34(15.89) 35(20.71) 1(2.50) 0 1(11.11) White 207(54.05) 113(52.80) 94(55.62) 32(80.00) 26(83.87) 6(66.67) Other 5(1.31) 2(0.93) 3(1.78) 0 0 0 Unknown 47(12.27) 32(14.95) 15(8.88) 4(10.00) 3(9.68) 1(11.11) Total 383 214 169 40 31 9

Clinical Success Yes 222(57.96) 124(57.94) 98(57.99) 23(57.50) 17(54.84) 6(66.67) No 53(13.84) 31(14.49) 22(13.02) 5(12.50) 3(9.68) 2(22.22) Missing or unknown 108(28.20) 59(27.57) 49(28.99) 12(30.00) 11(35.48) 1(11.11) Total 383 214 169 40 31 9

*If study therapy included a combination regimen, patients were included if the pathogens were resistant to either agent in the combination.

M.O. comment: Overall, the analyses based on treatment indication and demographics did not identify an etiology for why a higher proportion of cefepime patients with resistant pathogens at baseline died as compared with comparator patients. A higher percentage of cefepime patients [26.4% (19/72)] died from the “Mixed Infection” indication as compared with comparators [11.9% (5/42)]. The M.O. will discuss these patients in further detail in the following sections; however, based on a review of each patient’s CRF, the majority of these patients appeared to die due to complications from underlying co-morbidities.

The following two tables were prepared by Dr. Wu to further evaluate the 423 patients identified by BMS as having “resistant pathogens at baseline”.

197

Table 22. Mortality Status Stratified by Pathogen and Treatment Arm for 423 Patients with a “Resistant Pathogen at Baseline” Alive Death Total Cefepime Controls Total Cefepime Controls # of pts # of pts # of pts # of pts # of pts # of pts Pathogen name (multiple records per patient) C. Perfringens Hastiforme 1 0 1 Innocuum 1 0 1 Bacillus sp. 1 0 1 Propionibacterium sp. 1 1 0 Corynebacterium sp. 1 1 0 Jeikeium 1 0 1 Lactobacillus sp. 4 1 3 Bifidobacterium sp. 1 0 1 L. Monocytogenes 1 1 0 Faecalis 1 0 1 1 0 1 Strep. (Alpha hemolytic) 75 41 34 3 2 1 S. Viridans 3 0 3 Strep. (beta hemolytic) 2 1 1 Strep. (gamma hemolytic) 2 0 2 Strep. (Group D) 2 0 2 S. Agalactiae 9 7 2 1 1 0 S. Intermedius 1 1 0 1 0 1 Enterococcus 2 1 1 S. Bovis 49 26 23 5 4 1 S. Faecium 2 2 0 1 1 0 S. Salivarius 7 4 3 1 1 0 S. Mitis 3 2 1 S. Constellatus 7 6 1 1 1 0 S. Sanguis 1 1 0 S. Avium 4 1 3 S. Oralis 1 0 1 Staph. sp. (MS) 1 1 0 S. Aureus 1 1 0 Staph. Coagulase 33 10 23 6 4 2 S. Epidermidis 24 12 12 S. Saprophyticus 26 11 15 4 2 2 S. Hominis 1 1 0 S. Simulans (MS) 2 1 1 S. Haemolyticus 1 0 1 P. Acidilactici 7 5 2 1 1 0 S. Pneumoniae 1 0 1 Peptostreptococcus sp. 8 7 1 1 0 1 P. Anaerobius 3 3 0 P. Micros 2 2 0 Bacteroides sp. 1 1 0 B. Fragilis 4 3 1 B. Ovatus 13 6 7 B. Vulgatus 4 3 1 B. Thetaiotaomicron 5 5 0 B. Distasonis 6 2 4 B. Uniformis 10 7 3 P. Asaccharolytica 5 3 2 P. Intermedia 1 1 0 1 1 0

198

Alive Death Total Cefepime Controls Total Cefepime Controls # of pts # of pts # of pts # of pts # of pts # of pts Pathogen name (multiple records per patient) B. Capillosus 1 1 0 P. Disiens 1 1 0 P. Bivia(p+) 16 16 0 B. Ureolyticus 1 0 1 B. Stercoris 1 0 1 Pseudomonas sp. 3 0 3 P. Aeruginosa 34 19 15 2 2 0 P. Maltophilia 4 3 1 1 1 0 P. Cepacia 1 1 0 E. Coli 32 4 28 Citrobacter sp. 1 0 1 C. Freundll 2 0 2 S. Typhi 1 0 1 K. Pneumoniae 5 1 4 2 0 2 K. Oxytoca 2 0 2 E. Cloacae 9 2 7 E. Aerogenes 2 0 2 E. Intermedium 1 0 1 P. Mirabilis 2 0 2 H. Influenzae 1 0 1 H. Parainfluenzae (p-) 1 1 0 G. Vaginalis 4 3 1 P. Multocida 1 0 1 F. Necrophorum 1 1 0 F. Varium 1 0 1 Capnocytophaga sp. 1 1 0 C. Ochracea 1 1 0 Acinetobacter sp. 2 2 0 A. Anitratus 5 4 1 A. Hydrophila 1 0 1 Diphtheroids 1 0 1 M. Tuberculosis 1 1 0 Candida sp. 1 0 1 C. Albicans 1 1 0 Mucilaginosus 1 1 0 P. Stutzeri 1 1 0 Acinetobacter sp. 1 1 0 A. Baumannll 1 0 1 Total 485 246 239 35 23 12 *If study therapy included a combination regimen, pathogens were included if they were resistant to either agent in the combination. Patients were counted more than once if more than one pathogen was isolated at baseline.

M.O. comment: Overall, the number of patients with resistant pathogens at baseline who subsequently died was small. Regarding specific pathogens isolated at baseline in more than one patient in either study arm in this subset, more cefepime patients died post isolation of the following organisms versus comparator patients: Enterococcus sp. [cefepime: 9.0% (7/78) vs. comparator 3.2% (2/62)], S. aureus [cefepime: 28.6% (4/14) vs. comparator 8.0% (2/25)], S. epidermidis [cefepime: 15.4% (2/13) vs. comparator 11.8% (2/17)], and P. aeruginosa [cefepime: 9.5% (2/21) vs. comparator 0% (0/15)]. The M.O. notes that cefepime is not typically used for treatment of infections due to Enterococcus sp. or S. epidermidis. More control patients died post isolation of the following organisms versus cefepime patients: K. pneumoniae [cefepime: 0% (0/1) vs. comparator 33.3% (2/6)]. The M.O. will discuss these

199

patients in further detail in the following section; however, based on a review of each patient’s CRF, the majority of these patients appeared to die due to complications from underlying co- morbidities, and not due to the resistant pathogen isolated at baseline.

Table 23. Mortality Status Stratified by Source of Microbiologic Isolate for 423 Patients with a “Resistant Pathogen at Baseline” Source of Alive Death microbiologic isolate Total Cefepime Controls Total Cefepime Controls (multiple records per # of pts(%) # of pts(%) # of pts(%) # of pts(%) # of pts(%) # of pts(%) patients) Blood 122(25.15) 62(25.20) 60(25.10) 18(51.43) 11(47.83) 7(58.33) Intra-abdominal 109(22.47) 35(14.23) 74(30.96) 1(2.86) 1(4.35) 0(0) Sputum/lung 21(4.33) 9(3.66) 12(5.02) 5(14.29) 4(17.39) 1(8.33) Skin 2(0.41) 2(0.41) 0(0) 1(2.86) 1(4.35) 0(0) Urine 52(10.72) 25(10.16) 27(11.30) 3(8.57) 2(8.70) 1(8.33) Cerebrospinal fluid 1(0.21) 0(0) 1(0.42) 1(2.86) 0(0) 1(8.33) Other 139(28.66) 89(36.18) 50(20.92) 3(8.57) 3(13.04) 0(0.00) Unknown 39(8.04) 24(9.76) 15(6.28) 3(8.57) 1(4.35) 2(16.67) Total 485 246 239 35 23 12

*If study therapy included a combination regimen, pathogens were included if they were resistant to either agent in the combination. Patients were counted more than once if more than one pathogen was isolated at baseline.

M.O. comment: More patients with bacteremias died in both groups as compared to other sources for microbiologic isolates. This is not an unexpected finding.

The following table analyzed the relationship between age and mortality among the patients whom BMS noted had a resistant pathogen at baseline.

Table 24. Mortality Status Stratified by Treatment Arm versus Age for 423 Patients with a “Resistant Pathogen at Baseline” Age Alive Death Mean median range Mean median range Age 47.77(20.62) 48 0.17-96 62.68(19.02) 65.5 1.09-88 Cef 46.78(20.89) 46 1.00-92 63.61(16.98) 67 19-88 Con 49.02(20.27) 51 0.17-96 59.45(25.84) 63 1.09-87

M.O. comment: The patients with a resistant pathogen at baseline who died tended to be older in both groups. This is not an unexpected finding.

M.O. Death Narratives for the Forty Patients who Died within 30-Days of Study Therapy and were Identified by BMS as having had a “Resistant Pathogen at Baseline”

This M.O. reviewed the CRFs of the 40 patients reported by BMS as having at least one baseline pathogen resistant to study therapy and who died within 30 days post study therapy. BMS noted that if study therapy consisted of combination therapy, the pathogen was considered resistant if it was resistant to either drug in the combination.

200

Table 25. Forty Patients who Died within 30-Days of Study Therapy and were Identified by BMS as having had a “Resistant Pathogen at Baseline” Obs USUBJID INDICAT RANDTX DRUGNAME

1 411070002048 3 1 cefepime 2 411070043627 3 1 cefepime 3 411081005024 5 1 cefepime 4 411083051010 4 1 cefepime 5 411083051016 4 1 cefepime 6 411097014010 7 1 cefepime 7 411097026001 7 3 ceftazidime 8 411097043001 7 1 cefepime 9 411111003034 7 1 cefepime 10 411111004038 7 1 cefepime 11 411111004091 7 1 cefepime 12 411111008014 7 1 cefepime 13 411111008018 7 1 cefepime 14 411118001011 1 3 piperacillin, gentamicin 15 411126003049 7 3 cefotaxime, dexamethasone* 16 411160005005 10 1 cefepime 17 411160020054 10 3 ceftazidime 18 411160029031 10 1 cefepime 19 411160038005 10 1 cefepime 20 411186004021 1 2 cefepime, amikacin 21 411186011003 1 3 ceftazidime, amikacin 22 411186015004 1 2 cefepime, amikacin 23 411198002029 1 2 cefepime, vancomycin 24 411204010188 1 3 ceftazidime 25 411204016402 1 1 cefepime 26 411219014002 7 2 cefepime, amikacin 27 411219015005 7 2 cefepime, amikacin 28 411219015006 7 2 cefepime, amikacin 29 411219015025 7 3 ceftazidime, amikacin 30 411219021008 7 2 cefepime, amikacin 31 411219021009 7 2 cefepime, amikacin 32 411219021022 7 2 cefepime, amikacin 33 411219022002 7 2 cefepime, amikacin 34 411219022010 7 2 cefepime, amikacin 35 411219022011 7 2 cefepime, amikacin 36 411219024008 7 3 ceftazidime, amikacin 37 411219029008 7 3 ceftazidime, amikacin 38 411219030004 7 2 cefepime, amikacin 39 411219030018 7 2 cefepime, amikacin 40 411219032016 7 2 cefepime, amikacin *Patient 411126003049 was not included in the “Patient Level Analysis of Baseline Pathogen Susceptibility to Study Drug versus 30-Day Mortality Stratified on Treatment Group (Cefepime vs. Comparator)” because there was no comparable patient in the cefepime arm of the study which, in effect, made Study 411-126 a single-arm study. Therefore this patient was removed from this sub-group analysis. However, Patient 411126003049 was included in the overall analysis by Dr. Yu-te Wu. **RANTX=randomized treatment group; 1=cefepime monotherapy, 2=cefepime with adjunct therapy, 3=active comparator(s).

The key for the study treatment indications (INDCAT) follows: 1 = febrile neutropenia 2 = intra-abdominal infection 3 = lower respiratory tract infection 4 = urinary tract infection 5 = skin and skin structure infection 6 = bone and joint infection 7 = mixed bacterial infection 8 = endocarditis 10 = other

M.O. Death Narratives Patient AI411-070-002-048 was a 61 yo WM who was enrolled to receive cefepime for the treatment of a lower respiratory tract infection. The patient had lung cancer and emphysema

201

with a 3 week history of purulent sputum and possible new pulmonary infiltrate. Culture of the sputum grew P. aeruginosa resistant to cefepime. The patient was removed from the study after approximately 3 days of therapy because the treating physician felt that the patient did not have pneumonia. During the hospital course, the patient developed congestive heart failure with hypoxia. Two days prior to death he was diagnosed with a S. pneumoniae pneumonia and bacteremia. The patient was placed on piperacillin, clindamycin and gentamicin; however he died due to respiratory failure on (b) (6)

M.O. comment: The resistant pathogen isolated at baseline did not likely cause the patient’s death. Cefepime treatment failure was not the cause of the patient’s death.

Patient AI411-070-043-627 was a 72 yo WM who was enrolled to receive cefepime for the treatment of a lower respiratory tract infection. The CRF was in French. The patient had a history of prior pneumonia and cardiac insufficiency. Baseline blood cultures grew methicillin- resistant S. aureus (MRSA) and Enterococcus spp. both resistant to cefepime. The patient was switched to vancomycin and amikacin and follow-up blood cultures were negative. The patient died due to cardiac arrest and renal failure.

M.O. comment: The resistant pathogens isolated at baseline did not directly cause the patient’s death, but may have aggravated the patient’s underlying cardiac insufficiency which may have led to his renal failure. Cefepime treatment failure was not the cause of the patient’s death.

Patient AI411-081-005-024 was an 88 yo BF who was enrolled to receive cefepime for the treatment of a skin and skin structure infection. The patient had a prior history of atherosclerotic disease, stroke, decubitus ulcers, and urinary incontinence. She received 2 days of cefepime when the decubitus ulcer wound drainage culture grew MRSA, P. mirabilis (susceptible) and Group B Streptococci (susceptible); and blood cultures grew Enterococcus spp. and B. fragilis (susceptibility test not performed). The patient’s therapy was changed to vancomycin and metronidazole when culture results were obtained. The patient died due to multiple organ failure associated with sepsis 6 days post study therapy.

M.O. comment: The resistant pathogens isolated at baseline and underlying co-morbid conditions likely played a role in the patient’s death. It is less likely that cefepime treatment failure caused of the patient’s death.

Patient AI411-083-051-010 was a 75 yo WF who was enrolled to receive cefepime for the treatment of a urinary tract infection. The patient had a prior history of CAD, PVD, DM, right AKA, and decubitus ulcers. She received 3 days of cefepime when the baseline urine culture was reported as growing Enterococcus spp. The patient was switched to chloramphenicol. The family requested that the patient be “Do Not Resuscitate”, and she died due to cardiac arrest.

M.O. comment: It is unclear whether the resistant pathogen isolated at baseline contributed to the patient’s death. It is likely that her underlying co-morbid conditions played a role in her death. Cefepime treatment failure was not the cause of the patient’s death.

202

Patient AI411-083-051-016 was an 81 yo WF who was enrolled to receive cefepime for the treatment of a urinary tract infection. The patient was admitted with a GI bleed and had a prior history of HTN, CHF, severe COPD, organic brain syndrome, and renal calculi. Baseline urine culture grew Enterococcus spp. Blood culture was negative. The patient received 2 days of cefepime and then was switched to ampicillin. The patient had severe multi-system disease and the family requested that support be withdrawn. The patient died due to cardiopulmonary arrest, MI, and GI bleed.

M.O. comment: It is unclear whether the resistant pathogen isolated at baseline contributed to the patient’s death. Cefepime treatment failure was not the cause of the patient’s death.

Patient AI411-097-014-010 was a 78 yo WM who was enrolled to receive cefepime for the treatment of a “serious bacterial infection” (pneumonia). Medical history included: chronic HTN, aortic aneurysm and iliac aneurysm (s/p repair and bypass surgery (same admission), CVA x 3, COPD on mechanical ventilation, bleeding duodenal ulcer s/p exploratory laparotomy and truncal vagotomy (during same admission), and Alzheimer’s disease. Baseline tracheal aspirate cultures grew P. aeruginosa (resistant to cefepime) and Providencia stuartii. (Prior to enrollment, the patient had been on Ancef, gentamicin, vancomycin, and ceftazidime.) On Day 3 of cefepime, the patient was switched to piperacillin and tobramycin due to the resistant P. aeruginosa. During the hospital course, the patient developed aspiration pneumonia and went into multiple organ failure, the family withdrew support, and the patient died.

M.O. comment: It is unclear whether the resistant pathogen isolated at baseline contributed to the patient’s death. Cefepime treatment failure was not the cause of the patient’s death.

Patient AI411-097-026-001 was an 80 yo WM who was enrolled to receive ceftazidime for the treatment of a “serious bacterial infection” (UTI). Medical history included: MI with papillary muscle dysfunction (contemporary with study enrollment), atrial fibrillation, HTN, CHF, pulmonary edema, chronic renal insufficiency, Paget’s disease, s/p lacunar CVA with chronic disorientation. Prior to enrollment, the patient had been on ampicillin. Baseline urine culture grew E. aerogenes (sensitive to study drug), P. aeruginosa (sensitive to study drug), and Enterococcus spp. (resistant to study drug). The patient received 13 days of ceftazidime and ampicillin for the Enterococcus spp. Repeat urine cultures 6 and 8 days later were negative. The patient eventually died due to cardiopulmonary arrest and multiple organ failure.

M.O. comment: The resistant pathogen isolated at baseline did not likely cause the patient’s death given that post-baseline cultures were repeatedly negative. Ceftazidime treatment failure was not the cause of the patient’s death.

Patient AI411-097-043-001 was a 67 yo WM who was enrolled to receive cefepime for the treatment of “serious bacterial infections” (UTI, pneumonia, bacteremia). Medical history included: stage 4 non-Hodgkins lymphoma and a resolving middle lobe pneumonia. Baseline cultures included: arterial line blood culture grew E. coli (susceptible); central line blood culture grew S. haemolyticus (resistant); bronchial washings and tracheostomy culture grew Proteus vulgaris (susceptible); and urine grew K. oxytoca and K. pneumoniae (both susceptible). Repeat blood cultures 2 days later from arterial line and central line were negative followed thereafter

203 with intermittent culture positivity with coagulase-negative staphylococci (resistant). Within 3 days of starting cefepime, the patient appeared to be improving. He received a total of 10 days of cefepime; however, then the patient’s status worsened. Repeat tracheostomy culture grew P. (now Stenotrophomonas) maltophilia (resistant), the cefepime was stopped, and he was placed on Bactrim and ceftazidime. During the hospital course, the patient developed renal failure and reportedly expired due to acute renal failure, septicemia, respiratory failure, and progressive non- Hodgkins lymphoma six days after cefepime was discontinued.

M.O. comment: It is unclear if the baseline resistant pathogen, S. haemolyticus, contributed to the patient’s death. The patient’s underlying metastatic disease likely played a role in the patient’s death. It is unlikely that the patient’s death was due to cefepime treatment failure.

Patient AI411-111-003-034 was a 72 yo WF who was enrolled to receive cefepime for the treatment of a “serious bacterial infection” (pneumonia). Medical history included: paroxysmal tachycardia, HTN, GI adenocarcinoma with metastases to the lungs and liver, and pulmonary embolus. Baseline blood cultures were positive for Enterococcus spp. and S. bovis (both resistant to cefepime). The patient received 4 days of cefepime prior to being switched to “clamoxyl” [amoxicillin]. The hospital course was complicated by pulmonary embolus, ileus requiring surgical intervention, and CHF. Death was reported as due to metastatic adenocarcinoma, recurrent pulmonary emboli, and left ventricular failure.

M.O. comment: The patient’s death was most likely due to underlying co-morbid conditions. It is unlikely that the patient’s death was due to cefepime treatment failure.

Patient AI411-111-004-038 was a 19 yo WF who was enrolled to receive cefepime for the treatment of a “serious bacterial infection” (UTI). Medical history included: stage 4 ovarian sarcoma. Baseline urine culture grew E. coli (susceptible). Baseline cultures of pleural fluid and ascites grew E. faecalis (resistant). The patient received 13 days of cefepime for the E. coli UTI, in conjunction with ampicillin, Flagyl and vancomycin for peritonitis and empyema. The infections were reportedly controlled with the antibacterial therapy (repeat cultures negative); however the patient died due to inoperable, widely metastatic, Stage 4 ovarian cancer.

M.O. comment: The patient’s death was most likely due to underlying co-morbid conditions. It is unlikely that the patient’s death was due to cefepime treatment failure.

Patient AI411-111-004-091 was a 53 yo WM who was enrolled to receive cefepime for the treatment of a “mixed bacterial infection” (pneumonia). Medical history included: end-stage thyroid cancer, 25 yr smoking hx, and recent tracheostomy. Baseline sputum culture grew MRSA. Blood culture was negative. The patient died on the 2nd day of cefepime therapy. MRSA in the sputum persisted until death. The patient reportedly died due to underlying thyroid cancer.

M.O. comment: The patient’s death was most likely due to underlying co-morbid conditions. It is unclear if the MRSA was truly causing pneumonia or was a colonizing microbe. It is unlikely that the patient’s death was due to cefepime treatment failure.

204

Patient AI411-111-008-014 was a 64 yo WM who was enrolled to receive cefepime for the treatment of a “serious bacterial infection” (pneumonia). Medical history included: T.B. diagnosed in 1955, + tobacco history, persistent jaundice s/p cholecystectomy, and mechanical ventilator dependent. Prior to enrollment, the patient had received 3 days of cefotaxime for the suspected pneumonia. Baseline cultures revealed: Acinetobacter spp. (resistant to cefepime) from the sputum, throat, naval, and abdominal wound. Blood culture was negative, and repeat culture of abdominal wound drainage was negative on the 2nd day. The patient received 2 days of cefepime, and was concurrently on Flagyl. During the course of the hospitalization, the patient was diagnosed with pancreatic carcinoma and required a Whipple’s procedure. The patient died soon after surgery, reportedly due to underlying carcinoma, acute renal failure, and respiratory failure.

M.O. comment: The patient’s death was most likely due to underlying co-morbid conditions. It is unclear if the Acinetobacter spp. was truly causing a ventilator-associated pneumonia or was a colonizing microbe. It is unlikely that the patient’s death was due to cefepime treatment failure.

Patient AI411-111-008-018 was a 38 yo AM who was enrolled to receive cefepime for the treatment of a “serious bacterial infection” (infected ulcer). Medical history included: acute myeloid leukemia and cytotoxic cardiomyopathy. The patient had received amphotericin, Flagyl, amikacin, ceftazidime, “Azlocillin”, and flucloxacillin prior to enrollment for the scrotal ulcer (and possibly for empiric therapy for febrile neutropenia?). Baseline cultures revealed: Xanthomonas (now Stenotrophomonas) maltophilia (resistant to cefepime) from a scrotal ulcer. The patient received 4 days of cefepime, and then was switched to “Azlocillin”, vancomycin, and Augmentin for the resistant S. maltophilia. The patient reportedly died due to underlying acute leukemia unresponsive to chemotherapy and cardiomyopathy associated with chemotherapy.

M.O. comment: The patient’s death was most likely due to underlying co-morbid conditions: acute leukemia and cardiomyopathy. It is unclear if the S. maltophilia was associated with the patient’s death. It is unlikely that the patient’s death was due to cefepime treatment failure.

Patient AI411-118-001-011 was a 57 yo WF who was enrolled to receive piperacillin/gentamicin for the treatment of a febrile neutropenia. Medical history included: non-Hodgkin’s lymphoma (active) with an abdominal involvement, myelosuppression, and resected adenocarcinoma of the lung. The patient had received cefoxitin as perioperative prophylaxis. Baseline cultures included: 2 out of 4 blood cultures positive for two separate coagulase negative staphylococci (different susceptibility patterns). However, the patient was hypotensive and thought to be in septic shock. The patient received 4 days of gentamicin and 6 days of piperacillin along with concomitant vancomycin and Tazobactam (pharmacy error x 6 days). The patient was then placed on imipenem, aztreonam, and Flagyl. During the hospital course, the patient developed renal and respiratory failure, ascites, and had hyperbilirubinemia. No other pathogens were identified. The patient reportedly died due to cardiac arrest, multi-organ failure, and malignant lymphoma.

205

M.O. comment: The patient’s death was ultimately due to underlying malignant lymphoma. It is unclear if the coagulase-negative Staphylococci were associated with the patient’s death. It is unlikely that the patient’s death was due to study drug failure.

Patient AI411-126-003-049 was a 1 yo HM who was enrolled to receive cefotaxime for the treatment of bacterial meningitis. No additional past medical history was included in the CRF. Baseline cultures revealed: CSF and blood grew L. monocytogenes (resistant to study drug). The patient received 5 days of study drug, and then was switched to vancomycin and chloramphenicol. The hospital course was complicated with hypothermia, seizures, and coma. The patient died due to intracranial hypertension, cerebral edema, and bacterial meningitis.

M.O. comment: The patient’s death was due to bacterial meningitis caused by a pathogen inherently resistant to the study drug.

Patient AI411-160-005-005 was a 78 yo BM who was enrolled to receive cefepime for the treatment of “septicemia in hospitalized patients” (pneumonia). Medical history included: pneumonia requiring mechanical ventilation, HTN, GI bleed due to duodenal ulcer, anemia, traumatic fractures of right femur and left humerus, CVA resulting in left hemiparesis and seizure disorder, allergy to sulfa drugs, and elevated creatinine at baseline (1.8). The patient had received ceftazidime, Flagyl, and vancomycin for pneumonia within the week prior to enrollment into the study; and Flagyl and vancomycin were continued during the cefepime treatment period. Baseline cultures revealed: blood x 2 were negative; sputum was positive for MRSA; urine was positive for Candida spp. The hospital course was complicated by mental status change, dyspnea, tachypnea, hypotension, and hypothermia. The family requested comfort care only measures. Death was reportedly due to methicillin-resistant S. aureus pneumonia and cerebrovascular disease/insufficiency.

M.O. comment: The patient’s death appeared to be due to MRSA pneumonia (resistant to the study drug) and cerebrovascular insufficiency. The M.O. cannot rule out the possibility that cefepime may have contributed to the patient’s mental status changes.

Patient AI411-160-020-054 was an 81 yo BM who was enrolled to receive ceftazidime for the treatment of “septicemia in hospitalized patients” (bacteremia and UTI). Medical history included: cardiovascular disease, NIDDM, prostate cancer, PEG tube, foot ulcers, decubitus ulcer, hyperosmolar diabetic coma, dehydration, hypernatremia, and cachexia. Prior to enrollment, the patient received ceftazidime and gentamicin x 1 dose. Baseline cultures revealed: blood grew S. agalactiae, E. faecalis, and MRSA; urine grew E. coli. The hospital course was complicated by elevated AST (230) and ALT (199). The patient was reportedly discharged home with resolution of infections, but with some residual tachycardia and tachypnea. Three days later the patient experienced respiratory distress, was admitted to a non- study hospital, and died reportedly due to cardiovascular disease.

M.O. comment: The patient reportedly died due to cardiovascular disease. It is unclear if all of the patient’s infections resolved by discharge given that the MRSA, E. faecalis, and possibly the S. agalactiae bacteremias were not susceptible to the study therapy. Death did not appear to be due to study drug failure.

206

Patient AI411-160-029-031 was a 72 yo WF who was enrolled to receive cefepime for the treatment of “septicemia in hospitalized patients” (bacteremia). Medical history included: CAD, CHF, CVA with residual left sided weakness, COPD, 55 pack-yr tobacco history, respiratory failure prior to enrollment, cachexia, and penicillin allergy. Prior to enrollment, the patient received cephalexin x 1 dose, cefazolin x 2 doses, and ciprofloxacin x 2 doses. Baseline cultures revealed: blood x 3 grew Pseudomonas stutzeri (resistant to cefepime and ceftazidime). Repeat blood cultures performed 3-4 days later were reported as negative. The hospital course was complicated by progressive anorexia, poor PO intake, edema of hands reportedly due to very low serum albumin (1.7 gm/dL), and respiratory failure. The patient chose comfort care only measures, was discharged home, and reportedly died due to respiratory failure related to her underlying COPD.

M.O. comment: The patient reportedly died due to respiratory failure related to her severe COPD. The Pseudomonas stutzeri caused the patient’s bacteremia and was reported as resistant to cefepime. Death did not appear to be due to study drug failure.

Patient AI411160038005 was a 74 yo WM who was enrolled to receive cefepime for the treatment of “septicemia in hospitalized patients” (bacteremia). Medical history included: HTN, PVD s/p aorto-bifemoral bypass surgery, right BKA, carotid artery disease with mental status changes prior to enrollment, aortic stenosis, MI, renal stones, 120 pack-yr smoking hx, and a perforated duodenal ulcer. Prior to enrollment, the patient received vancomycin for 6 days, ceftazidime for 1 day, and Flagyl for 6 days. Vancomycin and Flagyl were continued during study therapy. Baseline cultures revealed: blood positive for E. coli x 3 bottles and E. faecalis (resistant to cefepime) x 1 bottle; urine grew E. coli, and sputum grew normal flora. Repeat blood cultures on the following day grew E. coli in 2 out of 6 bottles. Follow-up blood cultures thereafter were negative. During the course of the hospitalization, the patient developed an Acinetobacter baumanii pneumonia. The patient’s antibacterial therapy was switched to imipenem, and fluconazole was added for candiduria. Additionally, the hospital course was complicated by renal (creatinine=5.6) and hepatic failure (bilirubin=3.5), tachycardia, and depression. The family refused hemodialysis and the patient reportedly died due to cardiac arrest secondary to atherosclerotic heart disease and multiple organ failure.

M.O. comment: The patient reportedly died due to cardiac arrest and multiple organ failure. The patient was concomitantly prescribed vancomycin for the E. faecalis bacteremia. It appears less likely that the patient died due to study drug failure.

207

ondansetron, calcium, methylprednisolone, G-CSF, PRBCs, platelets, paracetamol, and vitamin B12. Vancomycin was added on 5/22. CXR on 5/24 revealed interstitial pneumonia. Additional antimicrobial therapy was added as follows: ciprofloxacin (5/25), amphotericin B (5/27), bactrim and ganciclovir (5/28), and imipenem (5/31). Amikacin was stopped on 5/28 and cefepime was stopped on 5/31. Repeat blood cultures from 5/21 to 6/2 were negative. Bronchoscopy on 6/2 was negative though CXR progressively demonstrated worsening diffuse pneumonia. Head CT on 6/6 showed multiple abscesses. The patient lapsed into a coma and died with respiratory (b) (6) (b) failure on days post cefepime and (6) days post amikacin). Autopsy revealed disseminated Aspergillus fumigatus infection (in the lungs and brain). The investigator deemed that the patient’s death was due to disseminated Aspergillus fumigatus infection. The following SAE was noted in the CRF: “diffuse Aspergillosis”(b) (6) was deemed unrelated to study therapy.

M.O. comment: This patient’s death was most likely due to disseminated Aspergillus fumigatus infection.

Patient AI411-186-011-003 was a 69 yo male (race unknown, from France) with non-Hodgkin’s lymphoma. Recent chemotherapy (12/17-12/18/92) included: cisplatin and cytarabine; he also received radiotherapy (8/10-8/28/92). On 12/28/92, he developed febrile neutropenia (Temp=38.5 C, ANC=750; however ANC fell to 85 on 12/31) and received ceftazidime 2gIVq8h and amikacin 500mgIVq12h for 8 days (12/28-1/4). Baseline blood cultures grew methicillin- resistant S. epidermidis (MRSE). No additional pertinent Med Hx was noted in the CRF. Pertinent concomitant therapy included: antibacterial agents as noted below, oral tobramycin and amphotericin B for gut decontamination; fraxiparine, G-CSF, lactulose, phenobarbitol, colchicine, PRBCs, platelets, dextropropoxyphene, and paracetamol. Vancomycin was added on 12/30 due to the MRSE positive blood cultures. On 1/1/93, the patient defervesced and his ANC=1688. On 1/4, ceftazidime and amikacin were discontinued due to completion of the course of therapy. On 1/6, the patient was noted to have purulent discharge around his catheter site that grew MRSE. Pefloxacin and fosfomycin were added. Repeat blood cultures from 1/8 were negative. Over the following week, the patient’s lymphoma progressed; his clinical status deteriorated, and was in a coma by 1/14. The patient died on (b) (6) days post ceftazidime and amikacin). No autopsy was performed. The investigator deemed that the patient’s death was due to progression of his underlying lymphoma. The following SAE was noted in the CRF: “death” (b) (6) was deemed unrelated to study therapy.

M.O. comment: This patient’s death was most likely due to progression on underlying lymphoma.

208

status worsened within hours of enrollment. Vancomycin was added on 6/16. Cefepime was discontinued on 6/17 after only 4 doses. The patient’s respiratory status progressively worsened culminating in ARDS requiring transfer to the ICU and intubation. Additional antimicrobial (b) agents were added: piperacillin, bactrim, ganciclovir, amphotericin B, and spiramycin. On (6) (b) (b) ( (6) days post cefepime and (6) days post amikacin), his pulmonary status worsened, he became anuric, and died. No autopsy was performed. The investigator deemed that the patient’s death was due to S. mitis sepsis leading to acute respiratory failure. The following SAE was noted in the CRF: “ARDS” (6/19(b) (6) was deemed unrelated to study therapy.

M.O. comment: This patient’s death was most likely due to S. mitis sepsis leading to ARDS and multiple organ failure. The S. mitis was reported as resistant to cefepime.

Patient AI411-198-002-029 was a 26 yo White female from Belgium with Hodgkin’s lymphoma. The patient received chemotherapy (starting after enrollment, 9/18/93); however the names of the agents were not included in the CRF. On 9/25/93, she developed febrile neutropenia (Temp=39.1 C, ANC=232) and received cefepime 2gIVq8h for 6 days (9/25-9/30) and vancomycin 1000-1500 mg/day for 5 days (9/25-9/29). Baseline blood cultures grew Stomatococcus mucilaginosus (resistant to cefepime, but susceptible to vancomycin). CXR was initially negative. No additional pertinent Med Hx was included in the CRF. Pertinent concomitant therapy included: Benzyl penicillin, ranitidine, G-CSF, morphine, epinephrine, dobutamine, dopamine, pancuronium, midazolam, and sufentanil. Repeat blood cultures from 9/26-9/28 were negative. On 9/26, the patient developed dyspnea and was found to have pneumonia on CXR. The patient progressed to ARDS, septic shock and multiple organ failure. (b) (6) (b) The patient died on (Day (6) of study therapy). Autopsy was performed; however results were not provided in the CRF. The investigator deemed that the patient’s death was due to pneumonia, ARDS, and septic shock. The following SAE was noted in the CRF: “ARDS” (9/26 (b) (6) was deemed to have an unknown relationship with study therapy.

209

electrolyte supplementation. Pertinent laboratory values: platelet count=26,000 at baseline (6/7), and 32,000 on 6/22/94. The patient defervesced, neutropenia resolved, repeat blood cultures were negative, and study therapy was stopped on 6/18. However, her overall status declined during the hospitalization and she elected for supportive care only on 6/23. The patient died on (b) (6) days post therapy). No autopsy was performed. The investigator deemed that the patient’s death was caused by progression of metastatic breast cancer. The following SAEs were noted in the CRF: mild vaginal bleeding (6/20 to death), mild skin breakdown of thighs and perineum (6/20 to death), and progression of breast cancer (7/1 to death). None of these SAEs were considered related to study therapy.

M.O. comment: This patient’s death was most likely due to progression of widely metastatic breast cancer. It is unclear what pathogen was resistant at baseline given that the K. pneumoniae was the only pathogen isolated at baseline and was susceptible to study drug.

Patient AI411-204-016-402 was a 43 yo White male from the USA with poorly differentiated nodular lymphoma. Recent chemotherapy included: fludarabine (4/93-7/93) and 2-CDA (8/93 through enrollment). On 9/13/93, he developed febrile neutropenia (Temp=39.2 C, ANC not performed; however WBC=0.3) and received cefepime 2gIVq8h for 4 days (9/13-9/16). One out of 4 baseline blood cultures was positive for E. faecium. Pertinent Med Hx included: macular rash on admission, herpes zoster infection 2 weeks prior to enrollment, dyspnea on exertion, pancytopenia, and tobacco use. Pertinent concomitant therapy included: antimicrobial agents as discussed below, acyclovir, G-CSF, hydrocortisone, benadryl, restoril, Tylenol, potassium and calcium supplementation, Ativan, PRBCs, and platelets. Pertinent laboratory results included: on 9/13 (creatinine=1.1, AST=733, ALT=846) and on 9/16-last day of recorded lab work- (creatinine=0.8, AST=270, ALT=486). On 9/16/93, the patient was discontinued from the study due to persistent fever and isolation of a resistant organism (E. faecium). His antibacterial therapy was changed to vancomycin, gentamicin, and cefoperazone. On 9/20, cefoperazone was discontinued due to suspected drug rash. Repeat CXR showed right middle and upper lobe infiltrates. Amphotericin was added empirically on 9/19. On 9/22, he was transferred to the ICU. On 9/24, open lung biopsy demonstrated Aspergillus pneumonia. Comfort care measures were initiated and the patient died on (b) (6) days post therapy). No autopsy was performed. The investigator deemed that the patient’s death was caused by persistent pancytopenia secondary to lymphoma and Aspergillus pneumonia and was not related to cefepime. The following was a notable SAE during the course of the study: sepsis exacerbation (9/16 to death) was deemed unrelated to cefepime.

M.O. comment: This patient’s death was most likely due to Aspergillus pneumonia, lymphoma, and pancytopenia.

Patient AI411-219-014-002 was an 85 yo WM who was enrolled to receive cefepime + amikacin for the treatment of “serious bacterial infections in hospitalized patients” (pneumonia). Medical history included: traumatic cervical injury resulting in quadriplegia, on mechanical ventilation, and + tobacco hx. Baseline cultures revealed: S. pneumoniae and S. agalactiae in sputum culture (both were sensitive to cefepime and resistant to amikacin). The patient reportedly improved on study therapy and was discharged from the ICU; however, the hospital course was complicated by what was believed to be a second episode of pneumonia due to A. baumanii (found in tracheal

210

secretions and resistant to cefepime). The patient was also noted to have one blood culture positive for a methicillin-resistant coagulase-negative staphylococcal species. The patient was initially placed on ceftazidime and then switched to piperacillin/tazobactam when susceptibility results returned. However, the patient was subsequently diagnosed with peritonitis thought to be due to either mesenteric ischemia or an intestinal perforation. The family elected not to proceed with further surgical intervention and the patient died on the following day.

M.O. comment: The patient died due to peritonitis. Though the S. pneumoniae and S. agalactiae isolated from the sputum were resistant to amikacin, they were both reported as susceptible to cefepime.

Patient AI411-219-015-005 was a 71 yo WF who was enrolled to receive cefepime + amikacin for the treatment of “serious bacterial infections in hospitalized patients” (pneumonia). Medical history included: treatment for a presumed pneumonia due to Xanthomonas (now Stenotrophomonas) maltophilia from approximately 2 weeks to within 1 day prior to enrollment into the study, asthma, hemorrhagic gastric ulcer that required a Billroth II procedure, and hip prosthesis. Prior to enrollment, the patient received ciprofloxacin and Bactrim for X. maltophilia pneumonia. Baseline cultures at enrollment revealed: P. aeruginosa and S. epidermidis from an endotracheal aspirate and protected bronchoscopic brush specimen. The hospital course was complicated by an intra-abdominal abscess (culture positive for E. faecium resistant to study therapy) in the area of the recent Billroth II procedure and another presumed pneumonia due to Xanthomonas (now Stenotrophomonas) maltophilia. The patient developed septic shock with hypotension, anuria, and multiple organ failure 2 days prior to death.

M.O. comment: The patient died due to intraabdominal sepsis related to a deep surgical site infection/abscess at the site of a prior Billroth II procedure. It is unclear if BMS designated the S. maltophilia or the S. epidermidis as the “baseline resistant pathogen”. The P. aeruginosa was reported as susceptible to study therapy.

Patient AI411-219-015-006 was a 79 yo WF who was enrolled to receive cefepime + amikacin for the treatment of “serious bacterial infections in hospitalized patients” (pneumonia). Medical history included: cardiac arrest due to community-acquired pneumonia resulting in anoxic encephalopathy, ventilator-dependent, mitral stenosis, and diabetes mellitus. Prior to enrollment, the patient received Augmentin for 8 days. Baseline cultures revealed: P aeruginosa (susceptible to study therapy) from endotracheal aspirate and protected brush specimen. However, repeat cultures were negative and patient reportedly completed therapy with improvement in respiratory symptoms. The hospital course was complicated by a UTI due to C. tropicalis which required therapy with fluconazole. The patient developed tracheal obstruction which along with the anoxic encephalopathy resulted in the patient’s death.

M.O. comment: The patient reportedly died due to tracheal obstruction and anoxic encephalopathy. It is unclear what baseline pathogen was resistant to study therapy. The P. aeruginosa was the only pathogen isolated at baseline and it was reported as susceptible to both cefepime and amikacin.

211

Patient AI411-219-015-025 was a 63 yo WM who was enrolled to receive ceftazidime + amikacin for the treatment of “serious bacterial infections in hospitalized patients” (pneumonia). Medical history included: hypertension, venous thrombosis, hepatic nodules, biliary dilatation, nephrolithiasis, hyperglycemia, suspected pancreatic cancer, and cerebral infarct with coma. Prior to enrollment, the patient received Augmentin for 3 days. Baseline cultures revealed: S. aureus (sensitive to cefepime and ceftazidime, but resistant to amikacin) from bronchial exudate culture. The hospital course was complicated by the confirmatory diagnosis of pancreatic adenocarcinoma from a liver biopsy specimen. The patient reportedly died due to metastatic pancreatic adenocarcinoma, cerebral infarct, and multi-organ failure, and not due to the S. aureus found in the bronchial exudate.

M.O. comment: The patient reportedly died due to metastatic pancreatic adenocarcinoma, cerebral infarct, and multi-organ failure. Thought the S. aureus was resistant to amikacin, it was reported as susceptible to ceftazidime.

Patient AI411-219-021-008 was a 49 yo WF who was enrolled to receive cefepime + amikacin for the treatment of “serious bacterial infections in hospitalized patients” (bronchitis, CXR negative for an infiltrate). Medical history included: meningioma and hemiplegia. Baseline cultures revealed: sputum positive for S. pneumoniae that was susceptible to cefepime and resistant to amikacin. During the hospital course, the patient developed a venous thrombosis that evolved into a pulmonary embolism and led to respiratory failure and death.

M.O. comment: The patient reportedly died due to respiratory failure from a pulmonary embolus.

Patient AI411-219-021-009 was a 58 yo WM who was enrolled to receive cefepime + amikacin for the treatment of “serious bacterial infections in hospitalized patients” (pneumonia). Medical history included that the patient presented for admission with a subarachnoid hemorrhage and in coma. Baseline sputum cultures were positive for P. mirabilis, S. pneumoniae, and K. pneumoniae (all were susceptible to cefepime; however, the S. pneumoniae was reported as resistant to amikacin). During the hospital course, the patient sustained a massive intracranial hemorrhage and died on the same day.

M.O. comment: The patient died due to a massive intracranial hemorrhage. All of the reported pathogens were susceptible to cefepime.

Patient AI411-219-021-022 was a 67 yo WM who was enrolled to receive cefepime + amikacin for the treatment of “serious bacterial infections in hospitalized patients” (UTI). Medical history included: hypertension, ulcerative colitis, “prostatic syndrome”, and “hemorragia digestiva”. Prior to enrollment, the patient received 2 days of ampicillin, clindamycin, and tobramycin as prophylaxis against peritonitis in the setting of “hemorragia gastrica”. Baseline cultures revealed: blood was negative; urine was positive for Enterococcus sp. (susceptible to cefepime and resistant to amikacin), Citrobacter sp. (susceptible to cefepime and amikacin), and Agrobacterium sp. (resistant to cefepime and amikacin). During the course of the study, the urinary pathogens were reportedly eradicated; however, the hospital course was complicated by

212

wound dehiscence s/p Billroth II procedure, enteric fistula formation, peritonitis, and post operative gastric hemorrhage that resulted in septic and hemorrhagic shock and death.

M.O. comment: The patient reportedly died due to septic and hemorrhagic shock associated with an intra-abdominal surgical complication.

Patient AI411-219-022-002 was a 59 yo WM who was enrolled to receive cefepime + amikacin for the treatment of “serious bacterial infections in hospitalized patients” (pneumonia). Medical history included: MI, COPD, diabetes, alcohol abuse, 1 ppd tobacco hx, and hyperlipidemia. Prior to enrollment, the patient received 5 days of Augmentin for “acute respiratory failure, intubation and mechanical ventilation”. Baseline cultures revealed: sputum positive for P. aeruginosa (susceptible to study therapy) and A. baumanii (borderline susceptible to cefepime and resistant to amikacin); blood culture was negative. The hospital course was complicated by congestive heart failure, pulmonary edema and suspected worsening pneumonia that resulted in the patient’s death after 9 days on study therapy. The investigator listed the death as caused by COPD, CHF, diabetes, and nosocomial pneumonia.

M.O. comment: The patient reportedly died due to CHF which may have been exacerbated by COPD and nosocomial pneumonia. The M.O. cannot rule out study therapy failure; however, it is noted that the A. baumanii was borderline susceptible to cefepime and resistant to amikacin.

Patient AI411-219-022-010 was a 73 yo WM who was enrolled to receive cefepime + amikacin for the treatment of “serious bacterial infections in hospitalized patients” (pneumonia). Medical history included: MI, COPD, hypoxic coma s/p MI, and on mechanical ventilation at the time of enrollment. Prior to enrollment, the patient received 14 days of erythromycin and “Cefonicid” for “community acquired pneumonia”. Baseline cultures revealed: bronchial secretion culture grew S. marcescens and K. pneumoniae (both reported as susceptible to study therapy); blood culture was negative. The hospital course was complicated by ongoing anoxic brain injury, suspected septic shock due to nosocomial pneumonia, and multiple organ failure. The patient reportedly died due to ischemic cardiomyopathy, COPD, anoxic brain injury, pneumonia, and multiple organ failure.

M.O. comment: The patient reportedly died due to multiple organ failure associated with ischemic cardiomyopathy, anoxic brain injury, and possibly nosocomial pneumonia. The M.O. cannot rule out study therapy failure. It is unclear what baseline pathogen BMS considered resistant to study therapy, because both S. marcescens and K. pneumoniae were reported as susceptible based on the CRF.

Patient AI411-219-022-011 was a 63 yo WM who was enrolled to receive cefepime + amikacin for the treatment of “serious bacterial infections in hospitalized patients” (pneumonia). Medical history included: duodenal ulcer, hypercholesterolemia, Guillain-Barre syndrome, chronic bronchitis, and on mechanical ventilation at the time of enrollment. Prior to enrollment, the patient received 4 days of ciprofloxacin as “empiric treatment (fever during mechanical ventilation)”. Baseline cultures revealed: MRSA in bronchial secretions. The patient received 2 days of study therapy and then was switched to vancomycin, imipenem, and non-study amikacin.

213

During the hospital course the patient developed renal insufficiency believed to be due to vancomycin. Approximately 21 days after the end of study therapy, the patient developed septic shock due to P. aeruginosa bacteremia and died 3 days later. The death was attributed to septic shock, ARDS, and Guillain-Barre syndrome.

M.O. comment: The patient reportedly died due to P. aeruginosa septic shock more than 20 days after the end of study therapy. The baseline resistant pathogen was MRSA.

Patient AI411-219-024-008 was an 86 yo WF who was enrolled to receive ceftazidime + amikacin for the treatment of “serious bacterial infections in hospitalized patients” (pneumonia). Medical history included: hypertensive cardiomyopathy, heart failure, chronic cerebral vasculopathy, and hyperuricemia. Prior to enrollment, the patient received 4 days of sulbactam/piperacillin for an unknown reason. Baseline cultures revealed: blood was positive for Acinetobacter baumanii (resistant to study therapy) and S. aureus (susceptibility not performed per investigator note); urine culture was negative. Despite this, the patient remained on study therapy for 9 days until death. The hospital course was complicated by renal failure thought to be due to amikacin and progressive heart failure. The patient reportedly died due to hypertensive cardiomyopathy and heart failure.

M.O. comment: The patient reportedly died due to hypertensive cardiomyopathy and heart failure. The A. baumanii was resistant to study therapy and the S. aureus was reportedly not tested for susceptibility, however, the patient was maintained on study therapy for 9 days until she died.

Patient AI411-219-029-008 was a 48 yo WM who was enrolled to receive ceftazidime + amikacin for the treatment of “serious bacterial infections in hospitalized patients” (pneumonia). Medical history included: recent intracerebral hemorrhage, + tobacco use, on mechanical ventilation. Prior to enrollment, the patient received. Baseline cultures revealed: S. pneumoniae (resistant to ceftazidime and amikacin) from the sputum; blood and urine cultures were negative. The patient received 5 days of study therapy. The hospital course was complicated by cerebral edema. The patient died due to cerebral hemorrhage and edema on the fifth day of study therapy.

M.O. comment: The patient reportedly died due to cerebral hemorrhage and edema. Even though the baseline pathogen was reported as resistant to study therapy, the investigator felt that the patient died due to his underlying co-morbid condition, i.e., cerebral hemorrhage.

Patient AI411-219-030-004 was a 56 yo WM who was enrolled to receive cefepime + amikacin for the treatment of “serious bacterial infections in hospitalized patients” (pneumonia). Medical history included: CNS bleed, HTN, and in coma. Baseline cultures revealed: P. aeruginosa (sensitive to cefepime and amikacin) in the blood and sputum. After completing a 21 day course of study therapy, the patient received ceftazidime and amikacin for an additional day prior to dying of intraventricular hemorrhage, intracerebral hematoma, and cerebral edema.

M.O. comment: The patient reportedly died due to complications related to his underlying CNS bleed. It is unclear what baseline pathogen BMS considered resistant to study therapy,

214

because the P. aeruginosa was reported as sensitive to cefepime and amikacin. The M.O. cannot rule out study drug failure.

Patient AI411-219-030-018 was an 88 yo WM who was enrolled to receive cefepime + amikacin for the treatment of “serious bacterial infections in hospitalized patients” (UTI and pneumonia). Medical history included: atrial fibrillation, subdural hematoma secondary to trauma, and coma. Baseline cultures revealed: urine positive for P. aeruginosa (sensitive to cefepime and amikacin); sputum positive for S. aureus (sensitive to cefepime and amikacin); and blood culture was negative. Repeat cultures of sputum and a culture of pleural fluid were negative. The hospital course was complicated by ongoing cerebral ischemia and seizures which began on Day 8 of study therapy and 3 days prior to death. The patient died due to traumatic head injury and cerebral ischemia.

M.O. comment: The patient reportedly died due to traumatic head injury and cerebral ischemia. The M.O. cannot rule out the possibility that the study therapy contributed to the onset of seizures; however, the patient also had head trauma at baseline.

Patient AI411-219-032-016 was a 64 yo WM who was enrolled to receive cefepime + amikacin for the treatment of “serious bacterial infections in hospitalized patients” (pneumonia). Medical history included: atrial fibrillation, COPD, laryngeal carcinoma, cardiac insufficiency, on mechanical ventilation, and + tobacco use. Prior to enrollment, the patient received Augmentin for 10 days for suspected respiratory infection. Baseline cultures revealed: bronchial secretions were positive for E. aerogenes (sensitive to cefepime and amikacin); and blood was negative. Study therapy was discontinued 1 week later when serology for Mycoplasma sp. and Legionella sp. returned positive. The patient was then switched to ceftazidime, erythromycin, and rifampicin. The patient died 4 days later due to heart failure, cardiac arrhythmia, COPD, and nosocomial pneumonia.

M.O. comment: The patient reportedly died due to heart failure, cardiac arrhythmia, COPD, and nosocomial pneumonia. The “nosocomial pneumonia” was apparently due to Mycoplasma sp. and Legionella sp. which would have been resistant to cefepime.

Additional analyses provided by Dr. Yu-te Wu

Figure 17. Patient Level Analysis of Patients with or without a Fungal Pathogen Isolated at Baseline versus 30-Day Mortality Stratified on Treatment Group (Cefepime vs. Comparator)

215

M.O. comment: The analysis by presence of a fungal pathogen at baseline did not demonstrate a statistically significant difference in 30-day mortality between cefepime and comparator regimens.

Figure 18. Patient Level Analysis of whether the Baseline Infection was Monomicrobial or Polymicrobial versus 30-Day Mortality Stratified on Treatment Group (Cefepime vs. Comparator)

M.O. comment: The analysis of monomicrobial versus polymicrobial baseline infection did not demonstrate a statistically significant difference in 30-day mortality between cefepime and comparator regimens.

Figure 19. Patient Level Analysis of Baseline Renal Failure versus 30-Day Mortality Stratified on Treatment Group (Cefepime vs. Comparator)

216

M.O. comment: The analysis of the presence of renal failure at baseline did not demonstrate a statistically significant difference in 30-day mortality between cefepime and comparator regimens.

Figure 20. Patient Level Analysis of Febrile Neutropenia Patients: Underlying Malignancy versus 30-Day Mortality Stratified on Treatment Group (Cefepime vs. Comparator)

M.O. comment: Overall, the analysis by malignancy type did not demonstrate a statistically significant difference in 30-day mortality between cefepime and comparator regimens. However, among the subset of patients with a solid tumor at baseline, 30-day mortality was higher among the cefepime patients. The level of confidence in this result is low given that the total number of patients who had a solid tumor was very small.

Figure 21. Patient Level Analysis of Febrile Neutropenia Patients: History of Prestudy Bone Marrow Transplant versus 30-Day Mortality Stratified on Treatment Group (Cefepime vs. Comparator)

217

M.O. comment: The analysis of history of bone marrow transplant at baseline did not demonstrate a statistically significant difference in 30-day mortality between cefepime and comparator regimens.

Dr. Wu also performed a Hazard Analysis of mortality stratified on treatment group.

Figure 22. Hazard Pattern of Mortality Stratified on Treatment Group (Cefepime vs. Comparator)

M.O. comment: Based on this analysis, the cumulative risk of death was similar between the cefepime and comparator patients.

Overall Conclusions of Yu-te Wu, Ph.D.:

218

Dr. Wu concluded in her meta-analysis that cefepime was not associated with a significantly higher mortality rate versus the comparator group, i.e., she did not confirm the findings of Yahav et al.1 Dr. Wu noted that her analysis was based on a broader spectrum of data sources and found that the mortality risk difference in the cefepime group was greater, but this increase was not found to be statistically significant in both trial-level and patient-level analyses. Additionally, she conducted subgroup analyses, most of which did not demonstrate a significantly increased mortality associated with cefepime treatment. However, significantly greater mortality was demonstrated for cefepime-treated patients in the following subgroups: patients treated for the indication of skin and skin structure infections, patients with baseline pathogens resistant to study therapy, and patients with febrile neutropenia with solid tumors at baseline. The numbers of subjects available in these subgroups were very small, thus resulting in wide confidence intervals. Therefore, significant findings from these subgroups would need to be re-examined if and when more data are available.

219

5 ADDITIONAL CLINICAL ISSUES

5.1 Sanz et al. (2002)

The following information was derived from one of the largest febrile neutropenia studies not directly sponsored by BMS.

Sanz MA, López J, Lahuerta JJ, Rovira M, Batlle M, Pérez C, Vázquez L, Julià A, Palau J, Gutiérrez M, Capote FJ, Ramos F, Benlloch L, Larrea L, Jarque I; Spanish PETHEMA Group. Cefepime plus amikacin versus piperacillin-tazobactam plus amikacin for initial antibiotic therapy in haematology patients with febrile neutropenia: results of an open, randomized, multicentre trial. J Antimicrob Chemother. 2002 Jul;50(1):79-88.

ABSTRACT FROM PUBMED: BACKGROUND: Standard therapy for suspected infections in patients with profound neutropenia is the combination of a beta-lactam antibiotic plus an aminoglycoside. Cefepime's broad-spectrum activity makes it an option for initial empirical therapy in neutropenic patients. The aim of this study is to evaluate the efficacy and safety of cefepime plus amikacin compared with piperacillin-tazobactam plus amikacin for initial empirical treatment of fever in adult haematology patients with severe neutropenia. METHODS: In this prospective multicentre trial, 969 patients with 984 febrile neutropenic episodes were randomized to receive iv amikacin (20 mg/kg every 24 h) combined with either cefepime (2 g every 8 h) or piperacillin-tazobactam (4 g/500 mg every 6 h). Clinical response was determined at 72 h and at completion of therapy. RESULTS: Eight hundred and sixty-seven episodes were assessable for efficacy (432 cefepime, 435 piperacillin-tazobactam). The frequency of success without modification of the empirical therapy was nearly identical for cefepime plus amikacin (49%) compared with piperacillin tazobactam plus amikacin (51%). Similar rates of success were found for microbiologically documented infection: 40% versus 39%, respectively. Antibiotic modification was necessary in 49% of cefepime and 44% of piperacillin-tazobactam patients. The overall response rate, with or without modification of the assigned treatment, was 94% in both groups. Drug-related adverse events were reported in 10% of cefepime plus amikacin versus 11% of piperacillin-tazobactam plus amikacin patients. Mortality due to infection occurred in a total of 10 patients (two cefepime, eight piperacillin-tazobactam). CONCLUSION: The empirical regimen of cefepime plus amikacin is equivalent to piperacillin-tazobactam plus amikacin in febrile adult haematology patients with severe neutropenia.

M.O. comment: This study was not included in the Yahav meta-analysis because deaths beyond those noted as “mortality due to infection” were not stratified on treatment group, i.e., it was one of the “missing” studies. Given the large number of febrile neutropenic patients enrolled in the Sanz et al. study, the Division was interested in obtaining additional information from the authors related to causes of death. After several attempts by both the Division and BMS, a colleague of Dr. Sanz provided the following table that contained information on 46 patients who died during the course of the study: 19 cefepime patients and 27 piperacillin/tazobactam patients.

220

Table 26. Patient-Level Mortality Data from Sanz et. al.17

221

UPN Regimen Date of End of the allocated Date of death Cause of death Isolated organism Susceptibility/ randomization regimen Resistance (b) (6) 14008 Cefepime 19/06/1998 02/07/1998 Superinfection None (clinical documentation) 19102 Cefepime 06/10/1999 18/10/1999 Initial infection Staphylococcus Amikacin R (cefepime epidermidis & pip/taz untested) 50005 Cefepime 18/09/1998 25/09/1998 Hemorrhage without infection ------69011 Cefepime 22/09/1999 04/10/1999 Other causes ------20006 Cefepime 10/06/1998 14/06/1998 Initial infection Pseudomonas aeruginosa Cefepime S, Pip/Taz R, Amikacin S 27004 Cefepime 14/08/1999 22/08/1999 Hemorrhage with infection Staphylococcus Cefepime S, Pip/Taz S, epidermidis Amikacin S 10023 Cefepime 05/09/1998 19/09/1998 Hemorrhage with infection Pseudomonas aeruginosa Cefepime untested, Pip/Taz S, Amikacin S 20089 Cefepime 16/02/2000 26/02/2000 Underlying disease progression ------without infection 20023 Cefepime 18/11/1998 01/12/1998 Hemorrhage with infection Streptococcus (viridans Untested group) 31009 Cefepime 09/06/1998 21/06/1998 Underlying disease progression Streptococcus (viridans Untested with infection group) 20025 Cefepime 18/11/1998 01/12/1998 Superinfection None (clinical documentation) 10082 Cefepime 06/04/1999 03/05/1999 Superinfection None (clinical documentation) 68009 Cefepime 28/12/1999 09/01/2000 Underlying disease progression None (clinical with infection documentation) 14036 Cefepime 29/01/1999 07/02/1999 Other causes ------20082 Cefepime 23/11/1999 29/11/1999 Underlying disease progression ------without infection 20074 Cefepime 25/09/1999 13/10/1999 Other causes ------50080 Cefepime 26/08/1999 02/09/1999 Underlying disease progression None (clinical with infection documentation) 33029 Cefepime 23/10/1998 30/10/1998 Underlying disease progression None (clinical with infection documentation) 20081 Pipertazo 21/11/1999 25/11/1999 Underlying disease progression ------without infection 19012 Pipertazo 10/07/1998 10/07/1998 Initial infection None (clinical documentation)

222

(b) (6) 12081 Pipertazo 31/12/1999 11/01/2000 Underlying disease progression ------without infection 19026 Pipertazo 27/08/1998 03/09/1998 Initial infection Streptococcus sanguis Amikacin R (cefepime & pip/taz untested) 33130 Pipertazo 23/01/2000 02/02/2000 Hemorrhage without infection ------33110 Pipertazo 30/12/1999 13/01/2000 Superinfection Candida krusei ------19008 Pipertazo 26/06/1998 03/07/1998 Hemorrhage with infection Klebsiella oxytoca Untested 20039 Pipertazo 24/03/1999 04/04/1999 Hemorrhage with infection Pseudomonas aeruginosa Cefepime I, Pip/Taz R, Amikacin S 10011 Pipertazo 05/07/1998 04/08/1998 Superinfection None (clinical documentation) 50006 Pipertazo 22/09/1998 28/09/1998 Initial infection Staphylococcus aureus Untested 31023 Pipertazo 12/08/1998 20/08/1998 Initial infection Staphylococcus Cefepime S, Pip/Taz S, epidermidis + Candida Amikacin S tropicalis 10039 Pipertazo 15/10/1998 04/11/1998 Hemorrhage with infection Stenotrophomonas Cefepime untested, maltophilia Pip/Taz R, Amikacin R 31024 Pipertazo 05/09/1998 15/09/1998 Initial infection Enterobacter cloacae Untested 31025 Pipertazo 10/09/1998 19/09/1998 Underlying disease progression ------without infection 19084 Pipertazo 08/07/1999 25/07/1999 Hemorrhage with infection None (clinical documentation) 33010 Pipertazo 19/11/1998 07/12/1998 Other causes ------12002 Pipertazo 15/06/1998 26/06/1998 Underlying disease progression None (clinical with infection documentation) 19006 Pipertazo 23/06/1998 17/07/1998 Initial infection None (clinical documentation) 20031 Pipertazo 05/01/1999 11/01/1999 Initial infection None (clinical documentation) 33092 Pipertazo 02/07/1999 12/07/1999 Superinfection Parainflueza virus ------50103 Pipertazo 02/12/1999 10/12/1999 Underlying disease progression None (clinical with infection documentation) 69013 Pipertazo 30/09/1999 07/10/1999 Initial infection None (clinical documentation) 19134 Pipertazo 23/02/2000 06/03/2000 Hemorrhage with infection None (clinical documentation) 10095 Pipertazo 14/05/1999 22/05/1999 Hemorrhage with infection None (clinical documentation)

223

(b) (6)

50011 Pipertazo 23/10/1998 08/11/1998 Underlying disease progression Coagulase-negative Untested with infection Staphylococcus 66006 Pipertazo 16/07/1999 25/07/1999 Underlying disease progression Acinetobacter baumannii Cefepime untested, with infection Pip/Taz R, Amikacin S 19118 Pipertazo 12/12/1999 22/12/1999 Other causes ------

M.O. comment: The M.O. created the following table to summarize the causes of death for patients in the Sanz et al. study.

Table 27. Summary of Causes of Death in the Febrile Neutropenia Study by Sanz et al. Cause of Death Cefepime Piperacillin/Tazobactam Initial infection 2 8 Superinfection 3 3 Hemorrhage without infection 1 1 Hemorrhage with infection 3 6 Underlying disease progression without infection 2 3 Underlying disease progression with infection 4 4 Other causes 3 2 TOTAL 18 27

Overall, the numbers of patients who died do to specific causes were similar in both groups, and do not appear to point to a specific safety signal associated with cefepime administration.

224

5.2 Current Cefepime Susceptibility

In addition, the Agency is currently investigating more recent data on cefepime susceptibility, particularly versus Pseudomonas aeruginosa and Enterobacteriaceae. Of note, EUCAST (European Committee on Antimicrobial Susceptibility Testing) under the auspices of the ESCMID (European Society for Clinical Microbiology and Infectious Diseases) has set the cefepime MIC for Enterobacteriaceae = 1. For Pseudomonas, the MIC = 8; however it specifically relates to the 2 gram IV every 8 hours dosing regimen.18 These breakpoints differ from FDA’s currently approved breakpoints.

M.O. comment: Dr. Frederic Marsik, Microbiology Team Leader and Dr. Charles Bonapace, Human Biopharmaceutics Team Leader, in conjunction with an SGE, will be further evaluating the cefepime breakpoints for Pseudomonas aeruginosa and Enterobacteriaceae.

225

5.3 Medical Officer Discussion

On May 16, 1997, the Agency approved cefepime for use as empiric monotherapy in febrile neutropenic patients. Cefepime is the only antibacterial agent approved for this indication and is currently listed as one of the first line empiric therapies for febrile neutropenia by the Infectious Diseases Society of America (IDSA).2 In May 2007, Yahav et al.1 published a meta-analysis in The Lancet Infectious Diseases that noted an increased 30-day mortality associated with the use of cefepime versus other β-lactams for the following combined clinical conditions: neutropenic fever, pneumonia, urinary tract/gynecologic infections, and other/mixed infections. However, upon further review of the paper by the Agency, the increased risk of death was only significant for the clinical condition of febrile neutropenia, and the authors were not able to provide a biologically plausible explanation for this apparent increased risk of death. This publication sparked further interest in a possible association between cefepime use and increased mortality. The Agency opened a dialogue with the holder of the reference listed drug (RLD), Bristol-Myers Squibb (BMS), in July 2007 to further investigate this issue. On November 14, 2007, the Agency released an “Early Communication” concerning the ongoing review of the reported increased mortality in patients who received cefepime,19 and on May 14, 2008, the Agency provided a “Follow-up to the November 14, 2007, Communication about the Ongoing Safety Review of Cefepime (marketed as Maxipime)”.20

Within the Agency, the analysis of mortality associated with cefepime versus comparable comparator agents was divided into several components.

1. The current M.O. reviewed the case report forms of all febrile neutropenic patients who died within 30 days of receiving cefepime or a comparator agent in the nine febrile neutropenia studies (7 comparative and 2 non-comparative studies) submitted to the US FDA by BMS. The M.O. attempted to identify the most likely cause of death, as well as, major factors (co-morbidities, adverse events, documented pathogens) that may have contributed to each patient’s death. Based on the clinical and nonclinical safety profile of cefepime, adverse events of special interest were identified and reviewed. These included any AEs associated with death, particularly neurological impairment/seizure, renal toxicity, liver toxicity, study drug failure, and central nervous system hemorrhage. The M.O. tallied this information and evaluated it for potential mortality trends.

As noted in Section 3.3 of this review, the current M.O. attempted to obtain information from BMS on the complete list of all clinical studies that included cefepime as a study agent. This encompassed all published and unpublished trials, as well as, all U.S. and non-U.S. trials including those not previously submitted to the FDA. Studies were characterized based on level of mortality data (patient vs. trial), whether mortality data were based on the Intent-to-Treat or the Clinically Evaluable population, whether mortality rates were based on actual patients vs. episodes of therapy (febrile neutropenia studies only), phase of study, study indication (clinical condition treated), comparator agent used (if applicable), combination regimen used (if applicable) use of blinding (if applicable), the country where the study was performed, duration of follow-up, and if the study was included in the Yahav et al. analysis. This process took over one year to

226

accomplish, included numerous information requests, and culminated in BMS’s submission of a revised “Master Table” of all cefepime clinical studies on 9/8/08. This information was eventually used, in part, to define the set of trials utilized by Yu-te Wu, Ph.D. from the Quantitative Safety and Pharmacoepidemiology group to perform her meta-analysis.

In addition, the Agency was able to obtain mortality data from an additional, large (969 patient) Spanish study that was originally noted by Yahav et al.1 and later by BMS as having missing mortality data. This study by Sanz et al.17 compared cefepime versus piperacillin/tazobactam for empiric therapy for febrile neutropenia. Sumati Nambiar, M.D., M.P.H., Deputy Division Director of Safety for DAIOP, was able to contact a colleague of Dr. Sanz who provided the missing mortality data from this large, comparative Febrile Neutropenia study from Spain. The M.O. analyzed the mortality information provided from the co-author for potential mortality trends.

2. The M.O. sought the data mining expertise of Ana Szarfman, M.D., Ph.D., to aid with the following: (1) to further characterize the study population that encompassed the nine febrile neutropenia studies; (2) to search for any additional deaths, beyond those originally noted by BMS, that may have occurred among the patients enrolled in the 9 febrile neutropenia studies (this was necessary because CDISC standards do not have a specific variable for death and BMS’s calculations of deaths varied from one submission to the next); and (3) perform additional analyses of patient deaths, including evaluating for AEs associated with the deaths, associations between patient co-morbid conditions and death, and the use of Bayesian meta-analytical methodology to assess potential causes of death (as developed by William DuMochel, Ph.D.).

3. The Quantitative Safety and Pharmacoepidemiology group was consulted to perform a new meta-analysis based not only on trial level mortality data from Yahav et al.1, but also to include mortality data from: (1) BMS studies originally submitted to the U.S. FDA that may not have been submitted for publication and were therefore omitted from the Yahav analysis, as well as, (2) unpublished cefepime studies that were never submitted to the U.S. FDA. In addition, the meta-analysis team reviewed some patient level data across multiple studies as per an agreed upon statistical analysis plan. This analysis was mainly performed by Yu-te Wu, Ph.D. Specifically, she performed the following analyses. • Dr. Wu repeated the Yahav et al.1 meta-analysis to ensure that the authors’ results were replicable. • She performed a trial level analysis of 30-day mortality on the 41 studies already referenced by Yahav et al.1 plus an additional 47 studies. These 88 studies included 9467 cefepime-treated patients and 8288 comparator-treated patients. • Dr. Wu performed a pre-specified analysis of patient level data on 35 studies. This third analysis included 5058 cefepime-treated patients and 3976 comparator- treated patients. Analyses of 30-day mortality were stratified based on the following variables: treatment indication (clinical condition treated), comparator drug, age, gender, race, location, any pathogen recovered at baseline, all pathogens isolated at baseline susceptible to study therapy, fungal pathogen recovered at baseline, baseline infection (mono- or polymicrobial), renal

227

[SEE THE SECOND ADDENDUM ATTACHED TO THIS REVIEW.] insufficiency or failure, active cancer or malignancy, and history of bone marrow transplant.

Thus far, no biologically plausible safety signal has been identified to explain the increased mortality observed in the meta-analysis by Yahav et al.1, either specifically among febrile neutropenia patients or including all cefepime patients studied in all known clinical trials for any clinical condition. The most straight-forward explanation for the mortality imbalance may be that appropriate randomization was not obtained and more severely ill patients were enrolled in the cefepime arm of some studies, in particular, some that studied febrile neutropenia. However, additional hypotheses were evaluated, including: (1) a search for previously unidentified adverse event(s), such as, a neurological effect leading to death from undiagnosed seizures/status epilepticus, and (2) a lack of microbiological efficacy. In addition, meta-analytic tools were utilized to attempt to identify any other possible cefepime-related adverse event(s) that might be linked to excess mortality.

The M.O. notes that only adequately powered and well-controlled prospective trials may definitively answer the question as to whether or not the use of cefepime is associated with increased mortality. However, given the practical limitations of executing such trials using a mortality endpoint, and in particular, additional trials to study cefepime as empiric monotherapy in febrile neutropenic patients, the following additional risk management activities will be recommended.

228

1. OSE, in conjunction with Premier Healthcare Informatics, is poised to perform a postmarketing analysis of mortality associated with the administration of cefepime versus comparable agents. 2. BMS, in conjunction with the (b) (4) Drug Utilization Database, is also poised to perform their own postmarketing analysis of mortality associated with the administration of cefepime versus comparable agents. 3. The Division may consider changing the dosing recommendations and/or mean inhibitory concentration (MIC) susceptibility breakpoints for P. aeruginosa and the Enterobacteriaceae pending further review by Clinical Microbiology and Human Biopharmaceutics. This inclusion of additional language in the “Microbiology” section of the U.S. label would improve consistency between the U.S. and EMEA product labels due to a change in bacterial resistance to cefepime (as well as other antibacterial agents). 4. The Division may consider providing additional clarification in the INDICATIONS/USAGE section of the label for “Empiric Therapy for Febrile Neutropenic Patients”. The proposed language is discussed in Section 6.4 of this review.

229

6 OVERALL ASSESSMENT

6.1 Conclusions

Cefepime is currently an important drug product in the armamentarium of antibacterial agents used to treat both Gram-positive and Gram-negative pathogens (including Pseudomonas aeruginosa and Enterobacteriaceae) that cause pneumonia, uncomplicated and complicated urinary tract infections, uncomplicated skin and skin structure infections, complicated intra- abdominal infections; and as empiric therapy for febrile neutropenic patients. As labeled, cefepime monotherapy is indicated for empiric treatment of febrile neutropenic patients. However, insufficient data exist to support the use of cefepime monotherapy in such patients at high risk for severe infection. The label delineates that patients at high risk for infection may include those with: (1) a history of recent bone marrow transplantation, (2) hypotension at presentation, (3) underlying hematologic malignancy, or (4) severe or prolonged neutropenia.

Based on the analyses performed by the Agency, cefepime administration was not associated with a statistically significant increased risk of death. Furthermore, no biologically plausible explanation for a potential increased risk of death associated with cefepime administration in febrile neutropenic patients was identified.

6.2 Recommendation on Regulatory Action

Due to its antibacterial activity against Gram-positive and Gram-negative pathogens, including Pseudomonas aeruginosa and Enterobacteriaceae; the lack of a clear, biologically plausible reason for the increased mortality observed in the meta-analysis by Yahav et al.;1 and the lack of a statistically significant mortality difference in the Agency’s meta-analysis (which included the studies in the Yahav meta-analysis plus additional studies not in the Yahav analysis), the M.O. currently recommends that cefepime remain on the market and maintain all approved indications. The M.O. notes that only adequately powered and well-controlled prospective trials may definitively answer the question as to whether or not the use of cefepime is associated with increased mortality. However, given the practical limitations of executing such trials using a mortality endpoint, and in particular, additional trials to study cefepime as empiric monotherapy in febrile neutropenic patients, the following additional risk management activities will be recommended.

6.3 Recommendation on Postmarketing Actions

6.3.1 RISK MANAGEMENT ACTIVITY

Both the Agency, through the Office of Surveillance and Epidemiology (OSE), and BMS will be performing additional postmarketing analyses of mortality associated with the use of cefepime

230

and comparable comparator agents using large hospital network databases. The M.O. recommends the following risk management activities. 1. OSE, in conjunction with Premier Healthcare Informatics, is poised to perform a postmarketing analysis of mortality associated with the administration of cefepime versus comparable agents. 2. BMS, in conjunction with the (b) (4) Drug Utilization Database, is also poised to perform their own postmarketing analysis of mortality associated with the administration of cefepime versus comparable agents. 3. Following the completion of an ongoing review, the Division may consider changing the dosing recommendations and/or mean inhibitory concentration (MIC) susceptibility breakpoints for P. aeruginosa and the Enterobacteriaceae. This inclusion of additional language in the “Microbiology” section of the U.S. label would improve consistency between the U.S. and EMEA product labels due to a change in bacterial resistance to cefepime (as well as other antibacterial agents). 4. The Division may consider providing additional clarification in the INDICATIONS/USAGE section of the label for “Empiric Therapy for Febrile Neutropenic Patients”. The proposed language is discussed in Section 6.4 of this review.

In the event that BMS chooses to perform additional prospective studies using cefepime in the future, both BMS and Agency should continue to evaluate for mortality imbalances against cefepime.

6.3.2 POST MARKETING REQUIREMENTS

6.3.3 OTHER PHASE 4 REQUESTS

The Agency asked BMS to perform their own meta-analysis to assess mortality associated with the use of cefepime versus comparable agents. (Results are found in Appendix 7.6 of this review.)

6.4 Labeling Review

The M.O. recommends that the following language be changed in INDICATIONS AND USAGE section of the label for the treatment indication: “Empiric Therapy for Febrile Neutropenic Patients”.

231

Original language from the INDICATIONS AND USAGE section:

Empiric Therapy for Febrile Neutropenic Patients. Cefepime is indicated for empiric monotherapy of febrile neutropenia. Antibiotic monotherapy may not be appropriate in patients at high risk for severe infection (including patients with a history of recent bone marrow transplantation, with hypotension at presentation, with an underlying hematologic malignancy, or severe or prolonged neutropenia). Insufficient data exist to demonstrate the efficacy of cefepime monotherapy in such patients (See CLINICAL STUDIES).

Proposed draft language for the INDICATIONS AND USAGE section:

(b) (4)

6.5 Comments to Applicant

1. The Agency recommends that BMS move forward with the proposed observational study of cefepime usage and mortality using the (b) (4) database. 2. The Division recommends the following change in the language for the INDICATIONS AND USAGE section of the cefepime label related to the “Empiric Therapy for Febrile Neutropenic Patients” indication. a. Original Language: i. “Empiric Therapy for Febrile Neutropenic Patients. Cefepime is indicated for empiric monotherapy of febrile neutropenia. Antibiotic monotherapy may not be appropriate in patients at high risk for severe infection (including patients with a history of recent bone marrow transplantation, with hypotension at presentation, with an underlying hematologic malignancy, or severe or prolonged neutropenia). Insufficient data exist to demonstrate the efficacy of cefepime monotherapy in such patients (See CLINICAL STUDIES).” b. Proposed Language: i. (b) (4)

232

7 APPENDICES

7.1 Regulatory Activity (Unabridged Version with M.O. comments)

June 30, 1992 Original NDA 50,679 was submitted. It contained two studies (AI411-118 and -131) in support of efficacy for use as empiric therapy in febrile neutropenia; and one additional noncomparative study (AI411-143) in support of safety for the proposed indication. Labeling proposed for this indication was as follows:

(b) (4)

Dr. Erhardt concluded that the data were inadequate to show that cefepime was equivalent to control regimens in empiric therapy of febrile neutropenic patients. Subgroup analyses suggested that cefepime monotherapy was inferior to control regimens in empiric therapy of febrile neutropenic patients. The data suggested that cefepime may have had a safety advantage over aminoglycosides. BMS was informed of the conclusions in a teleconference and responded that they considered these data to be an “interim report” and that they planned to submit a sNDA for this indication by the end of 1994.

233

complicated intra-abdominal infections, complicated skin and soft tissue infections, acute osteomyelitis, and acute bacterial arthritis or Appropriate results from two independent, adequate and well- controlled Febrile Neutropenia studies.

May 16, 1997 The Agency notified BMS that their supplemental application provided adequate evidence “for the empiric use of Maxipime® as monotherapy in febrile neutropenic patients”.

M.O. comment: The Yahav et al. meta-analysis will be discussed in further detail later in this review.

234

July 25, 2007 The Division requested that BMS provide interpretation and comments regarding Yahav et al.’s findings, as well as any necessary changes to the product labeling related to this issue. BMS responded on 8/16/07 that they disagreed with the authors’ conclusions citing incomplete mortality data and general problems associated with drawing conclusions based on a meta-analysis. BMS also stated that they performed an internal analysis and that mortality rates between cefepime and comparators for all indications excluding febrile neutropenia were similar, 5.5% (148/2707) vs. 6% (112/1842), respectively. All-cause mortality for febrile neutropenia was 8.1% (58/716) for cefepime and 6% (35/582) for all comparators. BMS concluded that no revision to the label was necessary based on the results of the meta-analysis.

August 30, 2007 BMS submitted initial analyses based on (1) mortality by indication and (2) mortality stratified by monotherapy versus combination therapy. These analyses were later revised and resubmitted by BMS on 10/5/07. Based on 10/5/07 analyses, BMS noted that a total of 317 patients died within 30 days of cefepime exposure in BMS sponsored clinical trials. The 30-day mortality rate for cefepime patients in comparative studies was 5.2% (203/3911) versus 5.8% (165/2840) for patients on comparator agents. Regarding mortality based on monotherapy versus combination therapy, BMS stated that their safety database contained “…195 reports with the event of death reported..” In 44.6% (87/195), cefepime was administered as monotherapy. In 55.4% (108/195) cefepime was administered as a component of combination therapy.

235

October 5, 2007 In addition to revisions to 8/30/07 responses, BMS provided summary data on neurological adverse reactions associated with cefepime. Based on cumulative review of spontaneous adverse event reports, the literature, and Phase 1-4 clinical trials, BMS identified 504 case reports that contained neurological adverse events. 470 were serious and 34 were non-serious. 364 were spontaneous in origin, 44 were from published literature, and 96 from Phase 1-4 clinical trials. The top five countries of origin for case reports were: Japan (97 reports), Switzerland (79), United States (74), France (68), and Belgium (51). Out of the 96 reports from clinical trials, 10 were considered likely/possibly related to cefepime. Of the 504 total reports, 123 resulted in death. Out of these 123 deaths, the neurological AE was reported as the cause of death in 20 cases: coma (4 reports), encephalopathy (3), cerebral hemorrhage (2), cerebral infarction (2), epilepsy (2), status epilepticus (2), and single occurrences of cerebral death, cerebral edema, cerebral ischemia, head injury, and neurological exam abnormal. 16 out of these 20 patients had concomitant medical history that may have contributed to the neurological AE (e.g., heart disease, malignancy, and sepsis).

236

AI411-168, Saito et al. 1992 (chronic respiratory tract infections)”

BMS responded on 12/16/07 and provided explicit mortality data for the three studies in question.

M.O. comment: As of December 2007, mortality data were not available for the following 7 studies. Chang, 1998 (severe infections; ceftazidime comparator (CFTZ))-36pts Chuang , 2002 (pediatric febrile neutropenia; CFTZ)-40pts Huang, 2002 (severe infections; CFTZ)-?52pts Jiang, 2003 (mod/severe lower resp tract infxns; CFTZ)-cannot find Kieft, 1994 (serious bacterial infxns/sepsis; (CFTZ)-133pts Lin, 2001 (adult pneumonia; (CFTZ)-41pts Sanz, 2002 (febrile neutropenia; cefepime+amikacin vs. CFTZ+amikacin) 969pts

November 14, 2007 The Agency released the following “Early Communication About an Ongoing Safety Review Cefepime (marketed as Maxipime)”.19 “An article in a recent issue of The Lancet Infectious Diseases has raised the question about increased mortality with the use of cefepime (Yahav D, Paul M, Fraser A et al. Efficacy and safety of cefepime: a systematic review and meta analysis. Lancet Infect Dis 2007; 7: 338–48). FDA is currently reviewing some safety data and has requested additional data to further evaluate the risk of death in patients treated with cefepime. Cefepime is a broad spectrum cephalosporin antibiotic currently approved for the treatment of a variety of infections due to susceptible strains of microorganisms. It is a member of the class of antibiotics known as β–lactams.

The article in the May 2007 issue of The Lancet Infectious Diseases describes a higher all-cause mortality in patients treated with cefepime compared to other β– lactams. Overall, the all-cause mortality was higher with cefepime than other β lactams (risk ratio [RR] 1·26 [95% CI 1·08–1·49]) and for the subgroup of patients with febrile neutropenia (RR 1·42 [95% CI 1·09–1·84]).

This early communication is in keeping with FDA’s commitment to inform the public about its ongoing safety reviews of drugs. FDA is working with the manufacturer of cefepime, Bristol-Meyers Squibb, to further evaluate the finding of increased mortality in patients who received cefepime. It will take about 4 months to complete this evaluation at which time FDA will communicate the conclusions and any resulting recommendations to the public. Until the evaluation is completed, healthcare providers who are considering the use of cefepime should be aware of the risks and benefits described in the prescribing information and the new information from this meta-analysis.

The FDA urges both healthcare professionals and patients to report side effects from the use of cefepime to the FDA's MedWatch Adverse Event Reporting program.”

237

On 1/31/08, BMS responded. BMS noted that the majority of the mortality cases (56/88) were from Febrile Neutropenia studies, and that the relationship of patient mortality to cefepime susceptibility among cultured pathogens could not be established in the absence of documented bacterial pathogens in more than half of the mortality cases. BMS also consulted (b) (4) M.D. Dr. (b) (4) noted that: “Literature reviews covering two publications critical to the cefepime breakpoint for susceptibility [Bhat, SV et al., 2007] or occurrences of “all-cause mortality” [Paul, M et al., 2006] show potential analysis flaws or confounding variables (cefepime dosing, patient morbidities, local resistance epidemiology, pathogen- specific outcomes, etc.) that effect results. Prospective examination of these parameters appear necessary and to re-examine the cefepime breakpoints as a component of harmonizing all broad-spectrum β-lactam interpretive criteria. These analyses should be comprehensive (penicillins, cephalosporins, carbapenems, inhibitor combinations, monobactams) and include simulation models via contemporary PK/PD applications.” “Generally, no in vitro susceptibility clues appeared as a factor for “all-cause mortality”, however a clear association of P. aeruginosa and Acinetobacter spp. treatment by lower dosing regimens was noted for the cases described Bhat et al. [2007]. Future considerations should address the possibility of an organism- specific breakpoint for these species correlating to the requirement for high-dose regimens and/or combination therapy.”

M.O. comment: The Division’s analysis and response to the suggestion of a need for organism-specific breakpoints and/or specified dosing regimens for particular pathogens, such as P. aeruginosa and Enterobacteriaceae will be discussed later in this review.

January 8, 2008 BMS in conjunction with the Swiss drug regulatory agency, Swissmedic, published an “Important Information for Health Care

238

Professionals” letter concerning, “Results from a meta-analysis publication reporting increased mortality associated with the use of cefepime.”

“Therefore, the Division requests that BMS obtain and provide the necessary mortality data (total number of patients in each treatment arm and the number of patients who died per treatment arm).”

The Division was able to make initial contact with Dr. Sanz via e- mail; however he subsequently did not respond to multiple requests for primary data on his study patients, e.g., causes of death for patients stratified by treatment group. Then, on 6/13/08, a colleague of Dr. Sanz, Dr. Isidro Jarque, responded with data on the causes of death for the clinically evaluable patients in the published febrile neutropenia study. Briefly, there were a total of 46 deaths out of 867 clinically evaluable patients (19/432 cefepime patients and 27/435 piperacillin/tazobactam patients).

M.O. comment: More detailed results from the Sanz et al. study will be provided in Section 5.1 of this review.

239

BMS provided the following response on 2/14/08: “The request received from the Agency specified studies that were conducted "since the initial approval of cefepime", and thus all studies that were conducted as a part of the initial filing were excluded. In addition, many of the post marketing trials for our antibiotic products that were not planned to support label change filings did not capture MIC or other susceptibility data. Thus, only those trials that were used to support label changes post initial approval were included in the data sent to Dr. (b) (4) . Using these selection criteria, the package sent to Dr. (b) (4) for his analysis was complete.”

BMS provided available CRFs in installments from 3/31/08 to 12/19/08.

240

provide pharmacokinetic parameters (e.g., plasma clearance, volume of distribution, elimination rate constant, and inter-compartmental constants) from healthy subjects from Phase 1 clinical trials using clinically-relevant doses, and 3) perform Monte Carlo simulations using the requested MIC and pharmacokinetic data to determine the probability of target attainment with a PK/PD target of 60-70% T>MIC for each organism listed under MIC Data. Further details are provided below.

MIC Data: Provide recent (within the past 3 years) MIC data for the organisms listed below. The data submitted should be the entire database of individual MIC values for each organism. At least 100 MIC values for each organism should be provided. This data should be used to perform the Monte Carlo simulations noted below.

Pseudomonas aeruginosa Escherichia coli Klebsiella pneumoniae Enterobacter cloacae Acinetobacter baumannii Acinetobacter species Serratia marcescens Enterobacter aerogenes Stenotrophomonas maltophilia Proteus mirabilis Klebsiella oxytoca Citrobacter freundii

When information is known as to the ESBL, Amp C, or KPC characteristic of each organism the MIC values should be provided separately for each of these characteristics as well as all MIC data combined together for each organism.

Monte Carlo Simulations: Monte Carlo simulations should be performed using the MIC data requested above and pharmacokinetic data to determine the probability of target attainment with a PK/PD target of 60-70% T>MIC for each organism listed under MIC data. Simulated plasma concentration-time profiles should be simulated for all approved dosage regimens (i.e., 500 mg q12h, 1000 mg q12h, 2000 mg q12h, and 2000 mg q8h). Simulated plasma concentrations should be corrected for the protein binding of cefepime (approximately 20%).”

BMS provided their response on 9/29/08. In the response, BMS noted that they did not feel that a change in breakpoints was justified.

M.O. comment: Dr. Charles Bonapace, Human Biopharmaceutics Team Leader, and Dr. Frederic Marsik, Clinical Microbiology Team Leader, reviewed this submission in detail. Further discussion of their analysis is found in their reviews as well as in Section 5.2 of this review.

M.O. comment: Up to this point, BMS still had not provided a complete list of all cefepime clinical trials conducted by BMS and non-BMS sponsors, design characteristics of the trials, or the mortality rates for cefepime and comparator. The purpose of the following information

241

request was to obtain information that would lead to a better understanding of the scope of all cefepime clinical trials.

March 17, 2008 The Division made the following information request. 7. Regarding the 10-5-07 corrected BMS response, Table 1, entitled, “Comparative Cefepime Trials by Indication”, describe the BMS-sponsored comparative studies by indication as noted in “Table 1” in further detail based on the following by trial: a. Name of comparator therapy b. Form of study blinding: i. Double-blinded ii. Outcome assessor blinded iii. Open label c. If combination study therapy, provide the full regimen

8. Regarding the 10-5-07 corrected BMS response, Table 2, entitled, “Non- Comparative Cefepime Trials by Indication”, subsection, “Multiple Bacterial Infections”, summarize the following by trial: a. Further delineate the types of infections treated/trial. b. Were these trials for compassionate use? c. Provide the reasons for the seemingly elevated mortality rates.

9. Regarding the 10-26-07 BMS submission (2,272 pages), “Attachment Q1.2: List of Unpublished Studies” (pages 334-342), please reformat the table to conform with the following tables from the 10-5-07 corrected BMS response, Table 1, entitled, “Comparative Cefepime Trials by Indication”, and, Table 2, entitled, “Non-Comparative Cefepime Trials by Indication”, providing, by indication, study number, and treatment arm the following: a. total number of subjects enrolled b. number and percent of subjects who experienced all-cause mortality within 30 days of cefepime therapy c. totals for the above per indication d. Additionally, provide the following further detail by trial i. Name of comparator therapy ii. Form of study blinding: 1. Double-blinded 2. Outcome assessor blinded 3. Open label iii. Treatment indication iv. If combination study therapy, provide the full regimen v. Country or countries where the trial was (were) performed

10. Regarding the 3-7-08 BMS submission, “Table 1: Sponsorship of Cefepime Trials Analyzed in Published Meta-analysis”, please provide the following: a. A complete list of all 57 studies cited by Yahav et al. (41 are accounted for in the 3-7-08 table) i. We understand that BMS responded to the sponsorship issue for the 16 missing studies in the 12-3-07 submission, however for completeness, and to have all of the studies/publications accounted for in one table, please include in this revised table. b. In the table please stratify by treatment indication and include: i. Yahav et al. reference number ii. primary author and year published iii. whether or not BMS sponsored the study and BMS study number

242

1. indicate whether the study was submitted to the U.S. FDA for registration purposes c. Regarding the Yahav et al. paper, Reference # 33, Gainer et al., 1995, Yahav et al. noted that the publication was missing mortality data, however, the 12-3-07 BMS submission on the 16 studies missing mortality data from Yahav et al. did not account for this publication. Please delineate whether or not this study was sponsored by BMS and incorporate the pertinent information into your response to this question. d. Regarding the 12-3-07 BMS submission, under “Table 1: Status and Disposition of 16 Published References in the Cefepime Meta-analysis Publication”, “Reference #41 (Huang et al., 2002)”, according to Yahav et al., the authors were able to obtain mortality data for Huang et al. (2002), however, the authors were not able to obtain mortality data for Huang et al. (2005). Please clarify whether BMS has mortality data for Yahav et al. Reference #42 (Huang et al. 2005) and incorporate the pertinent information into your response to this question.

11. Provide a list of all available studies for which BMS has access to primary data (both published and unpublished, submitted to the FDA and not submitted to the FDA). In the list, delineate which studies were double- blinded, outcome assessor blinded, and open label studies.

12. Please clarify if our understanding of the extent of data BMS has access to as outlined below is accurate: Of the 57 publications included in the Yahav et al. meta-analysis, 25 publications were based on BMS sponsored studies. Additionally, there are 68 BMS sponsored studies (comparative and non- comparative) that were not submitted to US FDA for registrational purposes. Please identify by study number which of the 68 “Unpublished Studies” were submitted to non-U.S. regulatory authorities. For each study, identify the non-U.S. regulatory agency to which BMS submitted the study. Some of these studies may have been included in the meta-analysis.

BMS provided their responses to these questions on 7/31/08.

Unless otherwise noted, all deaths are within 30 days of study drug discontinuation.

243

• In 5 studies, no report was available and all of the data were provided by the local BMS affiliate (AI411197, AI411247/P01-401, AI411168, AI411165/166, and AI411177). • In 10 studies, no report was available and the only data BMS has is from the CARES database (AI411222, AI411226, AI411228, CPM0895010, CPM099601, CPM2394006, CPM2298004, CPM 0895009, CPM484, and CPMUT97). BMS has searched local archives and to date, has not been able to find other documentation for these studies.

The initial list of 68 unpublished studies provided to FDA in the 26-Oct-07 response was reduced to 63 studies in the 13-May-08 response, and is being further reduced to 51 studies in this response. Table 1 outlines how the list of 68 studies was trimmed to 51.”

(b) (4)

M.O. comment: Additional analyses of the febrile neutropenia datasets were required to further evaluate the febrile neutropenia patients for potential reasons for the mortality imbalance.

Also on 3/17/08, the Division requested the following.

“We plan to expedite a careful re-evaluation of the conclusions for febrile neutropenia in the meta-analysis by Yahav et al by using processes that will facilitate the access and integration of the Sponsor’s raw clinical trial data collected for all the cefepime clinical trials that studied febrile neutropenia (including both those submitted to the US and the so called “Unpublished Studies”: 1. Comparative trials submitted to the US: a. AI411-131 b. AI411-189 c. AI411-204

244

d. AI411-118 e. AI411-137 f. AI411-186 g. AI411-198 2. Non-comparative trials submitted to the US: a. AI411-143 b. AI411-158 3. “Unpublished Studies” a. (b) (4) b. c. d. e. f. g. h. i.

Please plan to load, integrate, and validate the CDISC SDTM formatted clinical trial data collected for all the cefepime febrile neutropenia studies listed above before submission to the FDA. We plan to use a standard submission review tool to load and pool the validated data and integrated analytical tools to simplify the re-assessment of the cefepime data.

I. Please format the following datasets (domains) in the current version of CDISC/Study Data Tabulation Model (SDTM) version 1.1, SDTM Implementation Guide (SDTMIG) version 3.1.1.

Demographics = DM Concomitant medications = CM Exposure = EX Adverse events = AE

Use the same version of MedDRA across all trials.

Disposition = DS Medical history = MH Laboratory tests = LB

We are specifically interested in the following laboratory tests values

Serum chemistry: creatinine, alanine aminotransferase, aspartate aminotransferase, bilirubin, and alkaline phosphatase

Hematology: White blood cell count and absolute neutrophil count

II. Please create a Procedure dataset following the CM format described above.

We are specifically interested in bone marrow and organ transplant, dialysis, and autopsy report.

III. Format the following datasets (domains) using SDTMIG version 3.1.2.

Microbiology = MB Microbiology susceptibility = MS Pharmacokinetics Concentrations = PC

245

Pharmacokinetics Parameters = PP

IV. Please provide a text file that describes the indication of treatment, as follows:

The first variable of such a text file should contain the USUBJID. The second variable should contain the text of the indication. The third variable should contain a more comprehensive description of the indication and the criteria followed for evaluating the indication.

The three variables should be separated by a tab. The description text should not contain tabs or hard returns.

The first row should contain information for the first patient The second row for the second patient, etc

The following is a template format (see attachment):

indication_template .txt (186 B...

V. Please provide a text file with the integrated narrative for each patient who developed serious adverse event(s), withdrew due to an adverse event, and/or died. Please include the outcomes of every serious adverse event, including, but not limited to withdrawals and fatalities. Include all associated Causes of Death.

The first field of such a text file should contain the USUBJID. The second field should contain the text of the narrative. The USUBJID field and the narrative fields should be separated by a tab. The narrative text should contain information for all serious events, outcomes for serious adverse events, withdrawals, and deaths integrated into a single narrative text. The narrative text should not contain tabs or hard returns.

The first row should contain information for the first patient The second row for the second patient, etc

The following is a template format (see attachment):

narratives_templat e.txt (4 KB)...

VII. Additional Notes:

Please provide lab data using uniform units of measurement across all the trials.

Please follow the same character format for the USUBJID across all the trials and datasets, including narratives and indication text files.

Please follow the same CDISC format for dates across all the trials and datasets.

Please provide electronic copies of all protocols and protocol amendments.”

246

M.O. comment: BMS provided the requested datasets on 5/30/08; however, the following problems were noted which required resubmission of the datasets. 1. BMS could only provide patient level data for the 7 comparative and 2 non- comparative studies that were part of the Febrile Neutropenia sNDA package originally submitted to the U.S. FDA. BMS could not provide patient level data for any of the 9 “Unpublished Studies”. 2. Neither the CDISC format nor BMS provided a specific variable for deaths. 3. The ‘define.xml’ files were flawed. 4. BMS provided antimicrobial susceptibility data for the pathogens isolated from cefepime patients; however they did not provide antimicrobial susceptibility data for the pathogens isolated from comparator patients. Additionally, the CDISC format required separate domains for microbiologic specimens and microbiologic susceptibility data, and the domain for the microbiologic susceptibility data did not include a variable for the name of the pathogen making it difficult to link each pathogen to its respective antimicrobial susceptibility data, especially when multiple pathogens were isolated from the same specimen on the same date.

BMS provided revised Febrile Neutropenia datasets on 8/29/08. Further issues were noted. BMS changed some of the variable names making it difficult to incorporate data from the 5/30/08 submission with that of the 8/29/08 submission. However, Dr. Ana Szarfman was able to work around this issue successfully and link the appropriate data from one submission to the other.

On 11/14/08, BMS provided their response to this request and noted the following. “BMS contracted with (b) (4) to assemble the data and conduct the requested analyses. The report summarizing the information requested by FDA is provided in this submission.

The analysis and report of cefepime spectrum and activity is based on the SENTRY Antimicrobial Surveillance Program results for 1997 2007. The report includes 20 tables covering the organism groups listed by the USA FDA request. These data provide the requested MIC90s and MIC ranges, plus the MIC50 values, and demonstrate sustained activity against the indicated pathogens, but also note lesser activity against strains in some species (E. coli, K. pneumoniae) that have acquired ESBLs..

247

Bristol-Myers Squibb has reviewed the information and conclusions in Dr(b) (4) report, and concurs with his conclusions.”

M.O. comment: Dr. Frederic Marsik, Clinical Microbiology Team Leader, and Dr. Charles Bonapace, Human Biopharmaceutics Team Leader reviewed this submission in detail. Further discussion of their analysis is found in their reviews as well as in Section 5.2 of this review.

April 9, 2008 The Division and Sponsor held a teleconference. BMS agreed to: (1) attempt to contact the authors of the publications included in the meta-analysis that had missing mortality data; (2) provide further details about the 68 studies not submitted to the FDA (e.g., indications studied, mortality by treatment arm, and reasons for non-availability of data); (3) provide datasets for further analysis; and (4) conduct a meta-analysis that includes all publications that were included in the meta-analysis plus additional BMS sponsored studies that were not included in the meta-analysis.

On 5/13/08, BMS attempted to provide additional clarification on the 57 references in the meta-analysis publication and the 68 non- registrational trials, including: (1) numbers enrolled per arm, (2) number of deaths per arm, (3) extent of available individual patient data, (4) explanation for why data were unavailable, e.g., author not responsive or BMS data not retained.

The Division responded with the following additional information requests on 5/21/08. 3. With reference to the 5/13/08 submission to the Agency concerning "Clarification on the 57 references in the (Yahav et al.) meta-analysis and the 68 non-registrational studies described by BMS", provide further clarification related to your responses to Questions 1 and 2.

For each of the studies described in your responses to Questions 1 and 2, clarify whether the total patient numbers and deaths for cefepime and comparators were based on the Intent-to-Treat (ITT) population or the Clinically Evaluable population.

4. With reference to the 5/13/08 submission to the Agency provide further clarification related to your response to "Question 2":

On pages 5 and 6 of your response, you noted that 4 of the 68 "Unpublished" studies were originally listed twice and that one additional study, (b) (4) was incorrectly listed because it was not a study of cefepime, thus accounting for 63 "unpublished studies".

248

BMS provided their responses to the Division requests on 7/31/08.

M.O. comment: The submission on 7/31/08 was later revised by BMS and re-submitted on 9/8/08. Analysis of the submitted data is provided in the Safety section of this review.

May 14, 2008 The Agency provided a “Follow-up to the November 14, 2007, Communication about the Ongoing Safety Review of Cefepime (marketed as Maxipime)”.20 The communication noted the following.

May 27, 2008 The Division made the following three requests. Request #1: By 7/31/08, the Agency wants a complete accounting of all studies that dealt with cefepime, i.e., a "Master Table" with a complete listing and tally of all available studies with (1) mortality level data and (2) patient level data. All studies should be counted only once whether they were included in the Yahav paper or not, whether originally submitted to the US FDA for registrational purposes or not. In your accounting, indicate which studies where referenced in the Yahav paper and which studies where submitted to the US FDA for registrational purposes. In addition, indicate whether data are based on the Intent-to-Treat (ITT) or Clinically Evaluable (CE) population.

249

information using the variables STUDYID, and USUBJID and the same ID used in the patient-level dataset (PLD) and trial-level dataset (TLD) of April 30. Please refer to the following table on Quality Checks.

Request #3: Please reference the April 30, 2008 request for patient-level data, and the following BMS response sent in an e-mail to the Agency on 5/18/08:

"...Based on these discussions and Rho’s internal resource limitations, BMS will be able to deliver an integrated dataset (Tier I and Tier II) and trial-level data for the majority of the adult studies (26 of 31 studies) on 30-Jun-2008, with a cumulative integrated dataset containing the additional 5 adult studies delivered on 15-Jul-2008. An integrated dataset (Tier I and Tier II) for the 9 pediatric studies will be delivered on 31-Jul-2008."

Provide the integrated datasets and trial level data for all adult and pediatric studies in a single submission, please do not submit the information in a piece meal manner as was proposed above. Provide a date for when this will be accomplished.

BMS provided their responses to the Division requests on 7/31/08.

M.O. comment: The submission on 7/31/08 was later revised by BMS and re-submitted on 9/8/08. Analysis of the submitted data is provided in the Safety section of this review.

June 7, 2008 The Division made the following request. 1. Clarify whether the following three patients enrolled in Febrile Neutropenia studies died within 30 days of study drug completion. If these patients did die within 30 days of study drug completion, please provide their death narratives and case report forms. 411204008352 411186008009 411186012003

2. Please explain why the pediatric patients enrolled in Study AI411-131 were not represented in the Febrile Neutropenia datasets submitted to the Agency on 5-30-08.”

M.O. comment: BMS provided the requisite information on the 3 patients on 8/15/08. Per BMS, all three patients died > 30 days post last dose of study therapy.

On 8/29/08, BMS provided revised Febrile Neutropenia datasets that included the pediatric patients from Study 131. These datasets also included pathogen information with microbiologic susceptibility for both cefepime and comparator agents. However, BMS did not provide the death narratives for the pediatric patients in a text file format for integration into WebSDM. Additionally, the datasets still had problems (incorrect ‘define.xml’ files) that prevented appropriate integration of the data into the Agency’s Janus database.

June 9, 2008 The Division made the following request. 3. In addition to the 7/7/08 information request sent to BMS requesting clarification on whether three patients enrolled in Febrile Neutropenia studies (411204008352, 411186008009, and 411186012003) died within 30

250

days of study drug completion, please clarify whether the following additional three patients died within 30 days of study drug administration.

If these patients did die within 30 days of study drug completion, please provide relevant death narratives and case report forms. a. 411186036005 - This patient reportedly died due to the adverse event of "viral infection"; however the date of death was not provided. Last dose of cefepime appeared to be on 7/8/93. b. 411204005393 - This patient initially received cefepime from 9/2 9/13/93, and was listed as having been in the cefepime arm of the study; however the patient later received ceftazidime on 1/14/94 prior to dying on (b) (6) Was this patient re-enrolled in the study on 1/14/94 in the ceftazidime arm or was the patient only enrolled in the study once in 9/93? If the patient was re-enrolled in 1/94, provide a relevant CRF and Death Narrative for the 1/94 episode of febrile neutropenia. c. 411204016401 - This patient initially received cefepime from 9/4 9/10/93, and was listed as having been in the cefepime arm of the study; however the patient later received ceftazidime from 2/13-2/16/94 prior to dying on (b) (6) Was this patient re-enrolled in the study on 2/13/94 in the ceftazidime arm or was the patient only enrolled in the study once in 9/93? If the patient was re-enrolled in 2/94, provide a relevant CRF and Death Narrative for the 2/94 episode of febrile neutropenia. 2. Regarding the 5/30/08 submission of datasets for the 7 comparative Febrile Neutropenia studies, and specifically the “Microbiology Susceptibility Test” Domain Dataset for each respective study, we note that the MSRESCAT (Result Category) variable only contains interpretive susceptibility information (‘Susceptible’, ‘Intermediate’, ‘Resistant’, ‘Null’) against cefepime and no interpretive susceptibility information for the comparator study drugs used in the 7 respective studies.

M.O. comment: BMS provided the requisite information on the three patients in a submission on 9/2/08. Patient 411204005393 was later re-randomized to the ceftazidime group as 411204005471 and died <30 days post ceftazidime. Patient 411204016401 was later re- randomized to the ceftazidime group as 411204016484 and died <30 days post ceftazidime. Patient 411186036005 died > 30 days post therapy.

The revised Microbiologic Susceptibility data was provided on 8/29/08.

June 12, 2008 BMS provided an update on the status of their work trying to identify study and patient-level data from both the 57 studies cited in the Yahav et al. meta-analysis paper and 68 previously “Unpublished” studies (Phase 3 and 4 studies not previously submitted to the U.S. FDA).

251

July 15, 2008 The Agency and Sponsor held a teleconference to discuss data elements in the Febrile Neutropenia dataset submitted 5-30-08.

M.O. comment: The “Master Table” of all the cefepime clinical studies was later revised and resubmitted by BMS on 9/8/08. Analysis of the submitted data is provided in the Safety section of this review.

M.O. comment: Several key issues were noted at this point in the review. (1) Prior to this submission, both the Yahav paper and BMS had provided mortality data based on the arbitrary cut-off point of 30 days post therapy, therefore, to maintain consistency, the Division requested that BMS continue to provide data based on 30-day mortality. (2) In a paper by Cherif et al.,22 the authors indicated that 10 deaths occurred in the study. However, the “Master Table” provided by BMS indicated only 7 deaths in the study. Further investigation revealed that 7 patients died due to infections (3 in the cefepime arm and 4 in the imipenem arm) and three additional patients died due to underlying leukemia or lymphoma progression. It was not clear whether the three additional patients were in the cefepime or imipenem arms. (3) Many of the febrile neutropenia studies were complicated by the fact that patients could be randomized more than once if they sustained additional episodes of febrile neutropenia. In some of these situations, the investigators counted episodes of febrile neutropenia instead of patients. This complicated calculating 30-day mortality when the denominator included episodes instead of patients.

August 26, 2008 In a teleconference between the Agency and BMS, BMS confirmed that they were submitting both cefepime and comparator microbiologic susceptibility in the revised datasets due to be delivered on 8/29/08. BMS also agreed to submit an additional dataset that included both microbiologic susceptibility and patients’ treatment assignments.

252

M.O. comment: Treatment assignment was not originally included in the CDISC template for the Microbiology Susceptibility domain (dataset). It was determined that this information was necessary in the Microbiology Susceptibility domain because without it, several complex data transpositions would be required to determine treatment assignment for each patient. This could have resulted in errors that may have corrupted the integrity of the data; therefore it was important that BMS provide information on treatment assignment in the Microbiology Susceptibility domain.

BMS provided their responses on 9/10/08.

September 8, 2008 BMS submitted the revised “Master Table” of all clinical studies that contained a cefepime treatment arm.

September 25, 2008 BMS resubmitted the revised ‘define.xml’ files for the febrile neutropenia datasets.

September 29, 2008 BMS submitted their responses to Agency questions/comments related to cefepime MIC breakpoints and Monte Carlo simulations.

M.O. comment: Dr. Frederic Marsik, Clinical Microbiology Team Leader, and Dr. Charles Bonapace, Human Biopharmaceutics Team Leader, reviewed this submission in detail. Further discussion of their analysis is found in their reviews as well as in Section 5.2 of this review.

November 21, 2008 The Agency held a teleconference with BMS to further understand why more cefepime patients with resistant pathogens at baseline died as compared with comparator patients. BMS noted this may have be an artifact based on the fact that more comparator patients had missing information on baseline susceptibility to study therapy. In addition, BMS agreed to re-evaluate the integrity of their microbiology susceptibility datasets based on discrepancies noted by the Agency during their review of the originally submitted microbiology susceptibility data.

M.O. comment: On 12/12/08, BMS submitted a revised microbiology susceptibility dataset and noted that their primary analysis dataset was “…updated to accommodate dirty/missing data as revealed by investigation of differences between PRIMANAL [primary analysis] and MS/MB [microbiology] datasets…”

253

M.O. comment: On August 8, 2008, BMS submitted a proposal for a postmarketing analysis of mortality associated with the administration of cefepime versus comparable agents using the (b) (4) Drug Utilization Database. On 10/31/08, the Office of Surveillance and Epidemiology (OSE) provided comments/questions related to the proposed observational study. Briefly, OSE presented concerns related to BMS’s definition of febrile neutropenia, rationale for the proposed time period for data collection, inadequate sample size calculation, BMS’s mechanism for controlling for bias in the analysis, and lack of microbiological data in the data source. In addition, OSE requested a more detailed description of the proposed data source.

M.O. comment: It is likely that discussions between the Agency and BMS related to BMS’s proposed postmarketing study will continue after the current review is completed.

7.2 Highlights from the Original MO Review of BMS’s First Cefepime NDA (completed in 1994)

Three studies were submitted in the original NDA to support the indication of febrile neutropenia. Two studies were comparative: (1) Study 411-118 compared cefepime to the combination of piperacillin and gentamicin, and (2) Study 411-131 compared cefepime to ceftazidime. Study 411-143 was a non-comparative study.

The safety of cefepime in patients with fever and neutropenia was investigated in 156 patients involving 184 episodes of neutropenia (patients could be enrolled more than once) in three clinical trials. An additional 71 patients involving 74 episodes of neutropenia were treated with control agents. The M.O. reported that, “There was no significant difference in the mortality rate between cefepime and the comparators. There was also no significant difference in the incidence of clinical adverse events in the U.S comparative studies between cefepime and the comparators.”

“A significant difference was noted between cefepime and the comparator in the rate of discontinuation of therapy for adverse events. Patients in the cefepime groups were removed from therapy less frequently (p=0.02) than patients in the control groups for adverse events. The majority of the observed difference was derived from Study 411-118 where piperacillin/gentamicin was the comparator agent. In study 411 118 a large number of patients were removed from therapy for either clinical or laboratory evidence of renal toxicity.”

254

The percent of combined deaths in the two comparative studies were as follows: cefepime=10.5% (8/76) and control=6.8% (5/74).

M.O. comment: The original M.O. raised several concerns based upon a preliminary review of the febrile neutropenia data. (1) The safety and efficacy of cefepime for the treatment of febrile neutropenia had not been established in well controlled studies. Study 411-131enrolled only 22 patients in the cefepime arm (most from one study center) and “equivalence” with ceftazidime could not be demonstrated for the 95% confidence interval approach based on an intent-to-treat analysis. Study 411-118 did not demonstrate “equivalence” between cefepime and piperacillin/gentamicin. Study 411-143 was a non-comparative study. (2) Based on the limited number of patients enrolled in the comparative studies, clinical efficacy for cefepime monotherapy was lower than for control therapy: 78.6% (22/28) versus 96.4% (27/28), respectively. (In Study 411-118, clinical responses were 73% (16/22) for cefepime and 95% (20/21) for piperacillin/gentamicin. Per the original M.O.’s analysis, of Study 411-131, clinical responses were 100% (6/6) for cefepime and 100% (7/7) for ceftazidime.)

M.O. comment: The following was obtained from the Safety Summary of the clinical review of the original NDA, 1/4/94.

Cefepime at the time of NDA submission had been administered to 6,646 patients in clinical trials in North America, Europe and Japan. These patients were divided into 3 groups: 4,442 patients in Clinical Efficacy studies, 461 adult and 9 pediatric subjects in Pharmacokinetic studies, 4 cases of individual treatment use, and 1,730 Japanese patients as additional safety information. Of the patients in North America and Europe randomized to cefepime, there were 1,054 treatment courses with the zwitterion with sodium chloride formulation. The remaining treatment courses were administered with the dihydrochloride/arginine formulation. There were 1,842 patients who received control drug in the comparative studies out of which 79% (1,456) received ceftazidime.

M.O. comment: The original Medical Officer noted that it was often difficult and confusing to reconcile totals from subtotals in BMS’s tables because BMS was not consistent in their definition of “N”. In some instances “N” referred to the number of patients and in other cases “N” referred to the number of treatment courses. Additionally, “…the applicant often used the terms patient and treatment course interchangeably and often did not clarify which type of “N” was being employed.”

The confusion over use of numbers of patients versus numbers of febrile neutropenia episodes continues to be a problem today. In several instances, BMS was unable to determine the actual numbers of patients enrolled in some of the febrile neutropenia studies due to the fact that BMS and some investigators only tallied for episodes and not patients.

Regarding the review of safety data, the original M.O. reported that he had independently assessed the following: (1) case report forms of patients who died or discontinued treatment prematurely, and (2) raw lab data from the North American comparative trials provided by BMS in SAS files. The rest of the safety summary was based on the analyses submitted by BMS.

255

Deaths noted from the clinical review of the original NDA, 1/4/94 There were a total of 353 patients who died and whose CRFs were listed in the NDA. This included all patients enrolled in efficacy studies, all PK studies, all pediatric patients, and all compassionate use patients. Of the 353 patients who died, 350 deaths occurred in patients enrolled in clinical trials: cefepime=6% (244/4419) and control=6% (106/1842). The three additional deaths were as follows: two patients (2/4=50%) enrolled in compassionate use and one patient (1/470=0.2%) enrolled in a PK study.

Deaths Based on Region & Dose from the Original NDA Safety Summary (350 deaths in clinical studies) Drug North Europe Low Dose High America < 2g Dose > Q12h 2g q12h 138/3313 106/1106 83/2070 161/2349 Cefepime (4%) (10%) (4%) (7%) 42/359 28/640 78/1202 Control 64/1483 (4%) (12%) (4%) (6%) *Obtained from page 2 of original safety summary for NDA 50- 679 (no table number provided)

The original M.O noted the following based on the safety data from the original NDA and three subsequent safety updates.

“In the applicant discussion, the total number of patients are not given. However, the reviewer presumes that the total number of cefepime patients is 6646 (from the NDA) plus 742 (from the first and second safety updates) plus 967 (from the final safety update) or 8355 patients. If this is the correct total, then the overall death rate for cefepime would be 374/8355 = 4.5%. This is slightly less than the 247/4419 = 5.5% found in the original NDA.”

“Cefepime, itself, did not appear to be a direct toxin or cause of death. However, there were many deaths that were the result of failure or lack of efficacy of the product.”

M.O. comment: The M.O. noted that, “Although there were many deaths and many of these deaths were a result of drug failure, there were no deaths that appeared to be a direct, toxic consequence of the drug.”

7.3 Additional Information from the 1997 Joint Clinical and Statistical Review of Cefepime as Empiric Monotherapy for Febrile Neutropenia

SEVEN COMPARATIVE STUDIES Study 204: General information from JCSR Title: A Double-Blind Randomized Trial of Cefepime Versus Ceftazidime for the Empiric Treatment of Febrile Episodes in Neutropenic Cancer Patients Objective: To compare the clinical and microbiological efficacy and the safety of cefepime to ceftazidime in the treatment of febrile episodes in neutropenic cancer subjects. Investigators/Study Centers: See Table 204.2. Study design: This was a double-blind, randomized, comparative multicenter study conducted in the United States.

256

Subjects were stratified and separately randomized (1:1) based on their underlying malignancy; solid tumor or hematologic malignancy. Enrollment of 250 subjects was planned. Study Period: First subject enrolled January 16, 1993. Last subject completed therapy November 3, 1994.

M.O. comment: Study 204 was the only BMS-sponsored study for empiric therapy for febrile neutropenia submitted to the FDA for registrational purposes that was double-blinded. Therefore, the current M.O. reviewed this study in detail.

The following was noted on page 29 of the JCSR related to Study AI411-204. Duration of therapy: Study therapy was to be continued until the subject had been afebrile for 48 hours and recovered an absolute neutrophil count of greater than 500 cells/µL, or received 14 days of study therapy. Discontinuation of therapy: Study therapy could be discontinued early for any of the following reasons: An infection caused by a bacterial organism resistant to study therapy. A serious adverse event. Poor clinical response. Conditions requiring therapeutic intervention not permitted in the protocol. Personal preference of the subject or guardian. The investigator’s opinion that continuation of the study was not in the subject’s best interest. Occurrence of pregnancy. Subject lost to follow-up. All adverse events resulting in discontinuation of study drug were followed until resolution or stabilization. JCSR Medical Officer’s Comment Except for discontinuations due to adverse events, the protocol did not explicitly state how long patients who were discontinued from study therapy were followed. It was therefore not clear from the protocol if patients discontinued from the therapy were to be followed until resolution of fever or neutropenia had occurred, or until death occurred. However, the submission did generally contain follow-up data on such patients sufficient to allow determination of outcomes. Concomitant medications: Concomitant medications other than systemic antimicrobial agents were allowed as clinically indicated. These were recorded in the case report form with the indication, the dates started and discontinued, and the total daily dose. No other antibacterial agent was permitted during the study period except intravenous vancomycin. Vancomycin could be initiated for two reasons: 1) if fever persists for 72 hours after receipt of study drug and a gram-positive organism was suspected, or 2) if culture results demonstrated the presence of a methicillin-resistant staphylococci or enterococci at an infected site prior to 72 hours of study drug. Addition of vancomycin would be scored separately as a modification compared to other antibacterials. Antiviral and antifungal agents could be initiated and would be scored as a non-antibacterial modification. JCSR Medical Officer’s Comment As with all protocols except AI411-137, the protocol did not specify what specific modifications were to be made with respect to additional anti-bacterial agents.

The JCSR Medical Officer noted the following on page 30 related to Study AI411-204. The IDSA guidelines suggest that studies enrolling patients on prophylaxis should require that the same regimen be used for all subjects. Failing that, studies should either exclude or stratify patients on anti-microbial prophylaxis. In general, patients in this study were on a variety of regimens, but use of prophylaxis was not used as a stratum. Page 30-34 also noted the following concerning Study AI411-204. During Treatment Procedures: Procedures during treatment are summarized in Table 204.1. Maximal daily temperatures were recorded on the CRF. A temperature log was maintained by the subject if he/she was discharged home prior to end of treatment. The absolute neutrophil count was calculated daily. Seventy-two hours after study initiation, i.e., Study Day 4, the subject’s clinical status was evaluated by the investigator. This included a physical

257

examination with vital signs, assessment of the status of the infection, a blood culture as well as cultures of any site of focal infection. Serum chemistries were also repeated on this day. Adverse events, concomitant medications and administered blood products were recorded. Identical evaluations were performed every three days thereafter. After Day 4, cultures of blood and body sites were only obtained if clinically indicated. The results of all cultures were recorded in the case report form. All organisms considered pathogenic were identified and speciated to the extent possible. Positive cultures had susceptibility tests for cefepime and ceftazidime performed using the Kirby-Bauer disk method and/or an MIC panel when available. Zones of inhibition were recorded in the case report form, and the investigator judged whether isolates were causative pathogens or contaminants. JCSR Medical Officer’s Comment Except for coagulase-negative staphylococci, criteria for deciding whether organisms might be causative pathogens or contaminants were not explicitly stated in the protocol.

M.O. comment: The M.O. notes that the protocol lacked criteria for deciding whether organisms were considered pathogens versus contaminants.

Post-Treatment Procedures: Post-treatment procedures are summarized in Table 204.1. Subjects were to be evaluated at two time points after the last dose of study drug was administered. The first post-treatment evaluation occurred within 24 hours of completion of study drug. A physical examination and vital signs were performed. All subjects had blood cultures repeated unless a previous culture had been negative. Specimens from focal sites of infection were obtained if clinically indicated. Laboratory studies, CBC, and chemistries were repeated. Subjects were permitted to receive oral antimicrobials at the end of study therapy. This was recorded separately in the case report form. The second post-treatment evaluation occurred between the fourth and seventh post-therapy days. Discharged subjects were interviewed by phone to assess any intervening events, the presence of fever, and the use of any antibiotic therapy. Those subjects remaining hospitalized had a physical examination and vital signs performed. At this time the investigator categorized the subject’s clinical and bacteriologic outcome. JCSR Medical Officer’s Comment The IDSA guidelines call for a fixed follow-up period with a suggested length of 7 days. Sponsor’s Criteria for Evaluation Methods: Efficacy was determined by a consultant who had not participated in the study as an investigator, and who was blinded to patients’ treatment assignments. The consultant classified subjects into four diagnostic categories based upon clinical and microbiologic data - microbiologically documented infection, clinically documented infection, fever of uncertain origin, and non-infectious fever. Diagnoses: Infectious disease diagnoses were classified as:

Microbiologically Documented Infection (MDI): Bacteremia or fungemia involving one or more organisms without a definable non-hematogenous site of infection (primary) OR an infection at a specific site (e.g., UTI, cellulitis) that is microbiologically confirmed with or without bacteremia or fungemia.

Clinically Documented Infection (CDI): Signs and symptoms of infection at a specific site (e.g., UTI, cellulitis) but the microbial etiology could not be proven.

Fever of Uncertain Origin (FUO): Fever in the absence of localizing clinical signs and the microbial etiology of fever could not be proven.

Non-infectious fever: Fever in the absence of localizing clinical signs, no proven microbial etiology AND an alternative non-infectious cause is likely after thorough evaluation (e.g., tumor fever, drug fever).

Pathogens: All organisms obtained from cultures were classified by the investigators as causative, colonizer, contaminant, or normal flora and recorded in the case report form. Coagulase-negative staphylococci (CNS) required two or more separate blood cultures to be classified as a causative organism in bacteremia. CNS could be classified as a causative pathogen from a localized site only if it was the single organism isolated. Efficacy: Efficacy was evaluated on the basis of changes in signs and symptoms, of which temperature was the critical parameter. The efficacy evaluation also included an assessment of a microbial endpoint, when applicable. Three categories of clinical outcome were defined by the consultant: success, failure, and unevaluable.

258

Success. The subject’s fever and clinical signs of infection resolved, the infecting organism, whenever isolated, was eradicated without change in study therapy and the response was maintained for at least 4-7 days after discontinuation of study therapy. Failure. One of the following events occurred during or following therapy: - No response to study medication based on one of the following events: - Septic shock - Acute respiratory distress syndrome - Disseminated intravascular coagulation - Multiple organ failure - Progression of primary infection - Persistence of fever for 96 hours during study therapy - Pathogen resistant to study therapy - Persistent bacteremia (> 24 hours of study therapy) - Recurrent (breakthrough) bacteremia - Relapse of primary infection < 7 days post therapy - Death from primary infection Unevaluable. A subject was considered unevaluable in the following situations: - Initial infection caused by a virus, fungal, parasitic or mycobacterial organism. - A major protocol violation occurred, e.g. clinically inappropriate addition of a concomitant antibiotic, inadequate follow-up, or subject did not meet temperature or ANC criterion. - A non-infectious cause of fever was documented. - Early discontinuation of study therapy for an adverse event if the subject was clinically stable at the time of discontinuation but criteria for success or failure were not met. JCSR Medical Officer’s Comment As noted above, the protocol did not explicitly state whether patients who were discontinued from study therapy were to be followed until resolution of fever or neutropenia had occurred, or until death occurred, but such data were generally provided in the submission. In the modified intent-to-treat analysis, subjects with non-bacterial infections, adverse events, inadequate follow-up or those receiving concomitant antibiotics were all considered treatment failures. New infections were defined as infections, microbiologically or clinically documented, which had the onset of signs and symptoms during study therapy or during the follow-up period. A new infection could represent a breakthrough bacteremia with a new organism or an infection at a new site. A relapse or recurrent infection was defined as an infection with the same organism or worsening at the original site. New infections were recorded in the case report form and tabulated by treatment group. JCSR Medical Officer’s Comment Under the sponsor’s set of definitions, the occurrence of a new infection does not per se represent a treatment failure. The consultant then determined the evaluability of each case. Subjects were unevaluable if they did not meet a specified entry criteria (e.g., fever, neutropenia), have adequate post-treatment follow-up, or if the treatment regimen was modified within the first 72 hours or discontinued early without evidence for treatment failure. Evaluable subjects’ responses were then assessed as success or failure. A success was the defervescence of the subject on the initial treatment regimen without the addition of a new antibiotic, the eradication of the pathogen (if determined), the resolution of clinical signs and symptoms (if present), and the maintenance of this response through a 4-7 day period post-therapy. Reasons for treatment failure were specified, i.e., no response to treatment and persistent fever, resistant pathogen, progression of infection, relapse, or new infection. JCSR Medical Officer’s Comment In general, the Medical Officer’s evaluability and efficacy criteria were similar to those of the sponsor (see Introduction to the Reviews of Clinical Studies section). The following criteria for exclusion were substantially different from those of the sponsor: 1) modification prior to 72 hours for any reason other than isolation of a resistant

259

pathogen; 2) discontinuation at any point due to an adverse event; 3) absence of neutropenia within 48 hours of study entry. Episodes excluded for reasons 1) and 2) were included in the Medical Officer’s MITT analysis. Sponsor’s Safety Assessment All subjects receiving study therapy were evaluated for safety. All deaths, adverse events, and abnormal laboratory values were recorded. Their relationship to study therapy was assessed. Sponsor’s Statistical Methods All subject characteristics as well as description of antibiotic therapy were tabulated by treatment group for first course only. Median and range were used to describe continuous variables. Comparisons between treatment groups were done by the Cochran-Mantel-Haenszel test for categorical data and the two way ANOVA procedure based on ranks for continuous variables, both controlling for infection diagnoses. Success rates were analyzed for the first episode in the evaluable sample as well as in the modified Intent-to-Treat sample. The analyses of the difference in success rates were performed using the Cochran-Mantel-Haenszel test controlling for infection diagnoses. A stepwise logistic regression model was used to identify significant prognostic factors that may influence clinical outcome. The differences between the success rates for the treatment groups and its exact 95% confidence interval were reported. JCSR Medical Officer’s Comment The IDSA guidelines suggest evaluation by both patient and by episode (i.e., analysis of first episodes and all episodes). One objection to analysis of all episodes is that episodes occurring in the same patient are not independent events, given that the patient remains the same. Analysis of all episodes would then give undue weight to patients enrolled more than once. Following the IDSA guidelines, in the Medical Officer’s analysis, response rates were determined for both first episodes and all episodes. However, response rates for first episodes were used for the primary analysis.

M.O. comment: The M.O. notes that for the JCSR’s primary analysis, for patients that were enrolled more than once in the study, the response rate for the first episode of febrile neutropenia was used for the primary analysis.

Relevant Results for Study 204 According to the JCSR, page 36, Table 204.3, for enrolled patients, leukemia was the underlying malignancy in 29.4% (42/143) cefepime patients and 24.8% (33/133) ceftazidime patients. Solid tumors were the underlying cancer in 47.5% (68/143) cefepime patients and 51.1% (68/133) ceftazidime patients. The majority of patients had an absolute neutrophil count (ANC) < 500 of less than 7 days, 69.9% (100/143) for cefepime patients and 64.7% (86/133) for ceftazidime patients. Prophylactic antibiotics were administered to 20.3% (29/143) cefepime patients and 19.5% (26/133) ceftazidime patients.

M.O. comment: The M.O. notes that regarding underlying malignancy, more patients in the cefepime group had acute leukemia, and more patients in the ceftazidime group had solid tumors. This imbalance may, in part, have contributed to the increased mortality observed in the cefepime arm of Study 204.

Of note, the re-analysis of this demographic data by Dr. Ana Szarfman and the current M.O. did not reveal the imbalances noted by the JCSR. Per the re-analysis of Study 204, leukemia was the underlying malignancy in 22.1% (31/140) cefepime patients and 21.3% (29/136) ceftazidime patients. Solid tumors were the underlying cancer in 47.1% (66/140) cefepime patients and 49.3% (67/136) ceftazidime patients. Differences between the JCSR and current analyses may have been due to several reasons: (1) As discussed later in this review, several patients in Study 204 were randomized to alternate treatment regimens during the course of the study. For the purposes of the current 30-day mortality analysis, the current M.O. attributed the last randomized treatment regimen to the death. (The JCSR performed most of

260 their analyses based on study treatment received during the first episode of febrile neutropenia.) (2) The current demographic datasets did not contain categorical variables for “acute leukemia” or “solid tumor”, therefore, the current analysis team (Dr. Ana Szarfman and the current M.O.) created a customized algorithm for identifying acute leukemia and solid tumor patients (details of the algorithm are provided later in this review). If an algorithm was utilized by the JCSR team, it was not available for comparison. (3) BMS’s original datasets were not available. BMS reconstructed the original datasets and retrofitted the data into CDISC standard. It is possible that the reconstruction efforts, performed by BMS’s contractor, may have caused mistranslation of data as was noted in the Microbiology datasets (discussed later in this review).

The following was found on pages 37-38 of the JCSR.

Antimicrobial Prophylaxis: Prophylactic antimicrobial usage was infrequent and similar between the two groups (Table 204.4A). Seventeen percent of the population had received a systemic or non-systemic antibiotic within three days of study entry. Quinolone agents accounted for half of antibiotic prophylaxis. Eight percent of the study group received systemic antifungal prophylaxis, predominantly fluconazole. An additional ten percent were receiving non- systemic agents such as nystatin. Antiviral treatment with acyclovir had been received by seven percent of study subjects. Table 204.4A. Pretreatment Prophylaxis (First Episode)1

Number (%) of subjects

Prophylactic agent Cefepime Ceftazidime Total p-value (N = 143) (N = 133) (N = 276) Any prophylaxis 29 (20) 26 (20) 55 (20) 0.879

Antibacterial 26 (18) 20 (15) 46 (17)

Norfloxacin 8 (6) 6 (5) 14 (5)

Ciprofloxacin 8 (6) 4 (3) 12 (4)

TMP-SMX 2 (1) 2 (2) 4 (1)

Other 8 (6) 8 (6) 16 (6)

Antifungal 13 (9) 8 (6) 21 (8)

Fluconazole 12 (8) 8 (6) 20 (7)

Ketoconazole 1 (1) -- 1 (1)

Antiviral 8 (6) 10 (8) 18 (7)

Acyclovir 8 (6) 10 (8) 18 (7)

After the initiation of the empiric treatment, five subjects in the cefepime arm continued to receive oral antibiotics (ciprofloxacin (2 subjects), metronidazole, sulfacetamide, and oral vancomycin) as did two subjects in the ceftazidime arm (cephalexin and metronidazole). Antifungals and acyclovir were continued during at least part of the treatment phase in all subjects who received these agents pretreatment (Table 204.4B). JCSR Medical Officer’s Comment

1 Subjects may have received two or more classes of antimicrobial.

261

These antibiotics are all oral agents typically used for GI tract decontamination; although the efficacy of anti microbial prophylaxis in this setting has not been proven, their use has been relatively routine in the United States. JCSR Statistical Reviewer’s Comment There is no obvious imbalance between the treatment arms with respect to pretreatment prophylactic agents.

The following was obtained from page 39 of the JCSR and relates to patient evaluability in Study AI411-204.

Table 204.5. Episode evaluability 1° evaluability criteria MITT evaluability criteria Medical Officer Sponsor Medical Officer Sponsor All episodes 193/315 (61.3%) 219/315 (69.5%) 289/315 (91.7%) 295/315 (93.7%) Cefepime 103/163 (63.2%) 116/163 (71.2%) 151/163 (92.6%) 155/163 (95.1%) Ceftazidime 90/152 (59.2%) 103/152 (67.8%) 138/152 (90.8%) 140/152 (92.1%)

One hundred and twenty-two episodes (122) were excluded from the primary FDA analysis; 60/163 (36.8%) from the cefepime arm and 62/152 (40.8%) from the ceftazidime arm. Reasons for exclusion of episodes from the primary FDA analysis are shown in Table 204.6.

Table 204.6. Medical Officer’s reasons for exclusion from analysis Reason Overall Cefepime Ceftazidime Early modification 43/315 (13.6%) 20/163 (12.3%) 23/152 (15.1%) Non-bacterial infection 18/315 (5.7%) 11/163 (6.7%) 7/152 (4.6%) Not neutropenic 17/315 (5.4%) 7/163 (4.3%) 10/152 (6.6%) Lost to follow-up 13/315 (4.1%) 4/163 (2.5%) 9/152 (5.9%) Regimen D/C’d for ADR 12/315 (3.8%) 8/163 (4.9%) 4/152 (2.6%) Non-study Abx 10/315 (3.1%) 5/163 (3.1%) 5/152 (3.3%) Not febrile 6/315 (1.9%) 4/163 (2.5%) 2/152 (1.3%) Pre-existing infection 5/315 (1.6%) 4/163 (2.5%) 1/152 (0.7%) Non-infectious fever 3/315 (1.0%) 1/163 (0.6%) 2/152 (1.3%)

JCSR Medical Officer’s Comment Modification prior to 72 hours was the most common reason for unevaluability. This was done most often because the patient’s clinical status had deteriorated. Such patients were scored as failures by the sponsor. Although considered unevaluable under the primary analysis by the Medical Officer, these patients were treated as failures in the Medical Officer’s MITT analysis. JCSR Statistical Reviewer’s Comment The two treatment arms are balanced with respect to Medical Officer’s reasons for exclusion from analysis.

M.O. comment: More patients in the cefepime group were deemed nonevaluable due to study discontinuation because of an adverse drug reaction, 4.9% (8/163) for cefepime versus 2.6% (4/152) for ceftazidime. More patients in the ceftazidime group were deemed nonevaluable due to discontinuation because of early treatment modification, 12.3% (20/163) for cefepime versus 15.1% (23/151) for ceftazidime.

262

The following was obtained from pages 40-41 of the JCSR and relates to the infectious disease diagnoses observed in Study AI411-204. Infectious Disease Diagnoses The infectious disease diagnoses assigned by the Medical Officer and the sponsor for patients in the FDA evaluable, FDA MITT, and sponsor evaluable populations are shown in Tables 204.8A, 8B, and 8C, respectively. Table 204.8A. FDA infectious disease diagnoses for evaluable population Infection type Overall Cefepime Ceftazidime Any 193 (100%) 103 (100%) 90 (100%) MDI with bacteremia 36 (18.6%) 20 (12.3%) 16 (10.5%) MDI 18 (9.3%) 9 (5.5%) 9 (5.9%) CDI 10 (5.2%) 6 (3.7%) 4 (2.6%) FUO 129 (66.8%) 68 (41.7%) 61 (40.1%)

M.O. comment: Approximately one-third of the evaluable patient population (64/193) had evidence of a microbiologically or clinically documented infection. Out of which, 28% (54/193) had a microbiologically documented infection.

The following was obtained from pages 41-45 of the JCSR. Efficacy analysis Primary efficacy analysis: The primary efficacy variable was defined as the combined clinical and microbiologic response and was determined by the Medical Officer for each patient. The definitions of response are shown in Table 9.3A. The primary endpoint was outcome definition 1B applied to the evaluable population; for the MITT analysis, definition 1A was applied to the MITT population. Tables 204.9A, B, and C show response rates for all evaluable episodes as determined by the Medical Officer and by the sponsor; response rates for first episodes, and rates by treatment center, respectively. Because different definitions of outcome were applied to the FDA evaluable and MITT populations, the numerators differ between these two analyses.

263

Table 204.9A. All episode response rates Population Cefepime Ceftazidime 95% Confidence Interval 1 FDA evaluable 57/103 (55.3%) 56/90 (62.2%) 103, 90 (-0.2180, 0.0803) 55 3%, 62 2% 2 FDA MITT 37/151 (24.5%) 38/138 (27.5%) 151, 138 (-0.1386, 0.0779) 24 5%, 27 5%

Sponsor evaluable 65/116 (56.0%) 61/103 (59.2%) 116, 103 (-0.1721, 0.1083) 56 0% 59 2%

Sponsor MITT 65/155 (41.9%) 61/140 (43.6%) 155, 140 (-0.1362, 0.1035) 41 9% 43 6%

Table 204.9B. First episode response rates Population Cefepime Ceftazidime 95% Confidence Interval

FDA evaluable 49/88 (55.7%) 47/76 (61.8%) 88, 76 (-0.2245, 0.1013) 55 7%, 61 8%

FDA MITT 30/131 (22.9%) 33/119 (27.7%) 131, 119 (-0.1642, 0.0676) 22 9%, 27 7%

Sponsor evaluable 58/101 (57.4%) 52/87 (59.8%) 101, 87 (-0.1753, 0.1284) 57 4% 59 8%

Sponsor MITT 58/136 (42.6%) 52/121 (43.0%) 136, 121 (-0.1323, 0.1257) 42 6% 43 0%

The 95% confidence intervals are reported as nt,nc ( 95% C.I.) pt,pc where nt = number in the test group, nc = number in the control group, pt = response rate in the test group, pc = response rate in the control group. JCSR Statistical Reviewer’s Comment If all febrile neutropenic episodes in study AI411-204 are considered, cefepime fails to establish therapeutic equivalence to ceftazidime with respect to response rates for the patients who are FDA evaluable. Cefepime is therapeutically equivalent to ceftazidime with respect to response rates in patients included in the FDA MITT, sponsor evaluable and sponsor MITT populations. If the first febrile neutropenic episode is considered, cefepime fails to establish therapeutic equivalence to ceftazidime with respect to response rates for the patients who are FDA evaluable. Cefepime is therapeutically equivalent to ceftazidime with respect to first episode response rates in patients included in the FDA MITT, sponsor evaluable and sponsor MITT populations. JCSR Medical Officer’s Comment According to the DAIDP Points to Consider document, if the response rates for both test drug and comparator are below 80%, the 95% confidence interval around the difference in response rates should have a lower bound of no more than 20% for the drugs to be declared therapeutically equivalent. Under this criterion, the response rate obtained in the FDA analysis of the evaluable population does not demonstrate therapeutic equivalence. The differences between the Medical Officer’s analysis and the sponsor’s analysis result from the smaller size of the FDA evaluable population, as well as differences in assessment of outcomes. The sponsor supplied the Medical Officer with additional material derived from the case report forms; the response rates shown reflect review of this material in addition to the case report tabulations. Table 204.9C. Response rates by treatment center Cefepime Ceftazidime All centers 57/103 (63.2%) 56/90 (59.2%) 003 0/0 0/0 005 18/37 (48.6%) 18/32 (56.2%) 007 0/1 (0.0%) 0/0 008 3/3 (100.0%) 1/3 (33.3%) 010 12/15 (80.0%) 11/15 (73.3%)

1 Definition 1B was applied to the FDA evaluable population for the primary FDA analysis (clinical improvement and sustained defervescence achieved without modification of treatment (successful treatment of primary episode without new episode); completion of therapy with an oral antibiotic agent allowed. 2 Definition 1A was applied to the FDA MITT population for the main FDA MITT analysis (clinical improvement and sustained defervescence achieved without modification of treatment (successful treatment of primary episode without new episode); no post-therapy with oral antibiotic agents allowed.

264

012 2/5 (40.0%) 3/4 (75.0%) 013 3/3 (100.0%) 2/2 (100.0%) 014 8/11 (72.7%) 9/13 (69.2%) 016 10/22 (45.5%) 10/17 (58.8%) 017 1/2 (50.0%) 0/2 (0.0%) 018 0/4 (0.0%) 2/2 (100.0%)

Reasons for failure: A summary of the Medical Officer’s assessments of reasons for failure in evaluable patients is shown in Table 204.10.

Table 204.10. Reasons for treatment failure Reason for failure Cefepime Ceftazidime p value Persistent fever 18/103 (17.5%) 11/90 (12.2%) 0.483 Poor microbiologic response, initial 7/103 (7.0%) 5/90 (5.6%) isolate resistant Poor microbiologic response, initial 1/103 (1.0%) 0/90 (0.0%) isolate susceptible Death from primary infection 1/103 (1.0%) 2/90 Death from secondary infection 5/103 (4.9%) 0/90 (0.0%) Poor clinical response 7/103 (7.0%) 9/90 (10.0%) Bacteriologic relapse 0/103 (0.0%) 0/90 (0.0%) New MDI, susceptible isolate 0/103 (0.0%) 0/90 (0.0%) New MDI, resistant isolate 1/103 (1.0%) 4/90 (4.4%) New CDI 1/103 (1.0%) 0/90 (0.0%) New FUO 5/103 (4.9%) 3/90 (3.3%) Total failures 46/103 (44.7%) 34/90 (37.8%)

265

JCSR Medical Officer’s Comment Except for deaths due to secondary infection, there was not a significant difference between the treatment arms with respect to reasons for treatment failure. The deaths due to secondary infection in the cefepime arm were due to enterococcal bacteremia in two cases, an aspiration pneumonia in one case, and Aspergillus pneumonia in one case; the fifth death was attributed to infection but occurred in the setting of fever without a documented source of infection.

There was only one case in the cefepime arm in which failure occurred due to a poor microbiologic response with infection by a susceptible isolate, and none in the ceftazidime arm. There were no cases of bacteriologic relapse in either arm. JCSR Statistical Reviewer’s Comment The two treatment arms are balanced with respect to reasons for treatment failure. Microbiologic efficacy: Patients with microbiologically documented infections (MDIs) have the best documented evidence for having true bacterial infections. Response rates for MDIs in evaluable episodes as determined by the Medical Officer and the sponsor are shown in Table 204.11. Microbiologic efficacy by pathogen is analyzed in the integrated summary of efficacy (section 10).

Table 204.11. MDI response rates Population Cefepime Ceftazidime 95% Confidence Interval

FDA evaluable 9/29 (31.0%) 6/25 (24.0%) 29, 25 (-0.2043, 0.3450) 31%, 24% Exact 95% Confidence Interval 29, 25 (-0.2012, 0.3574) 31%, 24%

Sponsor evaluable 9/27 (33.3%) 4/22 (18.2%) 27, 22 (-0.1297, 0.4328) 33 3%, 18 2% Exact 95% Confidence Interval 27, 22 (-0.1328, 0.4449) 33 3%, 18 2%

JCSR Statistical Reviewer’s Comment The sample size of patients who had microbiologically documented infection is too small to ensure an acceptable power level for statistical inferences. However, based on the exact confidence intervals, cefepime fails to establish therapeutic equivalence with ceftazidime in patients with microbiologically documented infections included in the FDA evaluable population. The two treatment arms are therapeutically equivalent in the evaluable patient population defined by the sponsor.

M.O. comment: Though the proportions lacked statistical significance due to small numbers, in microbiologically documented infections (MDIs), cefepime-treated patients had similar if not slightly better response rates as compared with ceftazidime-treated patients.

The following was obtained from pages 46-47 of the JCSR and relates to treatment modifications and the frequency of superinfections or new febrile episodes in Study AI411-204.

Modifications A summary of the frequency of modification as determined by the Medical Officer, both overall and for specific classes of anti-microbial agents is presented in Table 204.13 for the evaluable AI411-204 population.

266

Table 204.13. Frequency of modification Cefepime (N=103) Ceftazidime (N=90) Overall (N=193) Any 65 (63.1%) 55 (61.1%) 120 (62.2%) Any anti-bacterial 59 (57.3%) 49 (54.4%) 108 (55.4%) vancomycin 27 (26.2%) 24 (26.7%) 51 (26.4%) aminoglycoside 12 (11.6%) 3 (3.3%) 15 (7.8%) cephalosporin 27 (26.2%) 12 (13.3%) 39 (20.2%) ß-lactam/penem/mono-bactam 16 (15.5%) 20 (22.2%) 36 (18.7%) quinolone 12 (11.6%) 8 (8.9%) 20 (10.4%) metronidazole 3 (3.3%) 1 (1.1%) 4 (2.1%) anaerobic coverage 15 (14.6%) 7 (7.8%) 22 (11.4%) Anti-fungal 22 (21.4%) 18 (20.0%) 40 (20.7%) Anti-viral 12 (11.6%) 9 (10.0%) 21 (10.9%) Mean time to modification (d) 4.78 ± 3.36 4.27 ± 2.98 4.55 ± 3.17

JCSR Medical Officer’s Comment It is difficult to interpret differences in the frequency of modification by anti-bacterial class, since they may have been affected by local practice patterns and generally occur in the absence of microbiologic data. Superinfections or new febrile episodes A summary of the frequency of new febrile episodes or documented infections in evaluable patients is presented in Table 204.14. There did not appear to be a significant difference between the treatment arms. Interestingly, the majority of new episodes in both arms were due to fever without a clinical or microbiologic source. Table 204.14. Frequency of new febrile episodes or infection Nature of 2nd event Cefepime (N=103) Ceftazidime (N=90) Overall (N=193) All 14 (13.6%) 8 (8.9%) 22 (11.4%) MDI (same isolate) 0 (0.0%) 0 (0.0%) 0 (0.0%) MDI (new pathogen) 5 (4.9%) 3 (3.3%) 8 (4.1%) susceptible 0 (0.0%) 0 (0.0%) 0 (0.0%) resistant 5 (4.9%) 4 (4.4%) 9 (4.7%) CDI 1 (1.0%) 0 (0.0%) 1 (0.5%) FUO 8 (7.8%) 4 (4.4%) 12 (5.2%)

267

Table 204.8B. FDA infectious disease diagnoses for MITT population Infection type Overall Cefepime Ceftazidime Any 289 (100%) 151 (100%) 138 (100%) MDI with bacteremia 59 (20.4%) 32 (21.2%) 27 (19.6%) MDI 27 (9.3%) 11 (7.3%) 16 (11.6%) CDI 20 (6.9%) 12 (7.9%) 8 (5.8%) FUO 183 (63.3%) 96 (63.6%) 87 (63.0%) Table 204.8C. Sponsor’s infectious disease diagnoses for evaluable population Infection type Overall Cefepime Ceftazidime Any 219 (100%) 116 (100%) 103 (100%) MDI with bacteremia 36 (16.4%) 20 (17.2%) 16 (15.5%) MDI 13 (5.9%) 7 (6.0%) 6 (5.8%) CDI 11 (5.0%) 5 (4.3%) 6 (5.8%) FUO 159 (72.6%) 84 (72.4%) 75 (72.8%)

JCSR Medical Officer’s Comment These results demonstrate that the majority of infections in both arms were due to fever without microbiologic or clinical evidence of infection.

M.O. comment: The JCSR M.O. noted that the majority (approximately two-thirds) of patients in the study had fever without microbiologic or clinical evidence of infection.

The following was obtained from pages 48-49 of the JCSR and relate to the safety analysis of Study AI411-204.

Safety analysis Mortality: Thirty-two subjects died in association with the first episode of febrile neutropenia during the course of the trial, for an overall mortality rate of 11.6%. There were 22 deaths in the cefepime treatment group and 10 in the ceftazidime group. An analysis of deaths by specific cause is shown in Table 204.15. Table 204.15 Cause of Subjects’ Deaths by Investigator - First Episode

Cefepime Ceftazidime Total p-value1 (N = 143) (N = 133) (N = 276) Total deaths 22 (15.4%) 10 (7.5%) 32 (11.6%) 0.038

Initial infection 3 (2.1%) 3 (2.3%) 6 (2.2%) 1.00

Secondary infection 9 (6.3%) 2 (1.5%) 11 (4.0%) 0.062

Underlying disease 10 (13.3%) 5 (3.8%) 15 (5.4%) 0.293

Two-thirds of deaths occurred eight days or more after initiation of empiric therapy in both arms; time to death and degree of neutropenia did not differ significantly between groups. Examination of cases of death due to secondary infection in the cefepime arm revealed two cases of bacteremic superinfection with Enterococcus faecium, one each of superinfection with Staphylococcus epidermidis, Clostridium septicum, and a Bacillus species, one case of candidemia, one case of Aspergillus pneumonia, one case of aspiration pneumonia, and two cases of fever of uncertain origin. In the ceftazidime arm, one superinfection death was due to candidemia and one to fever of uncertain origin. JCSR Medical Officer’s Comment

1 Fisher’s exact test

268

Some of these deaths occurred well after the study drug had been stopped; in particular, the death due to C. septicum infection occurred sufficiently long after study drug discontinuation that the patient involved was scored as a treatment success. There does not appear to be a discernible pattern to the deaths by secondary infection in either arm. JCSR Statistical Reviewer’s Comment There was a statistically significant difference in total deaths due to a higher mortality rate with cefepime. This appears to be due to deaths from secondary infections or underlying disease.

M.O. comment: The M.O. notes that at the time of the original review of Study 204, the higher mortality rate in the cefepime group was attributed to deaths from secondary infections and underlying disease. According to the JCSR, time to death and degree of neutropenia did not differ significantly between the two study arms.

Discontinuations due to adverse events: Reasons for discontinuation of study therapy due to adverse events are shown in Table 204.16. There was no significant difference between treatment arms.

269

Table 204.16 Discontinuation of Study Therapy Due to an Adverse Event, First Episode Number (%) of Subjects

Cefepime Ceftazidime Total (N = 143) (N = 133) (N = 276) Any Adverse Event 10 (7) 6 (5) 16 (6)

Rash 7 (5) 2 (2) 9 (3)

Sepsis 1 (1) 2 (2) 3 (1)

Hypotension 1 (1) 1 (1) 2 (1)

Fever and Rash -- 1 (1) 1 (1)

Epiglottitis 1 (1) -- 1 (1)

Cochran-Mantel-Haenszel p-value = 0.293

JCSR Statistical Reviewer’s Comment The two treatment arms are balanced with respect to reasons for discontinuation due to adverse events. Clinical adverse events: Adverse events were generally evenly distributed between the treatment groups. Nausea was the most frequent event with 28 reported events occurring with cefepime and 27 with ceftazidime. Diarrhea was more frequent in the ceftazidime group compared to cefepime (16% versus 23%). Other important differences between the treatment groups were noted for the incidence of abdominal pain (17% versus 10%) and hypotension (15% versus 3%).

M.O. comment: Though it was unclear from the JCSR M.O.’s statement, based on analysis of the original AE dataset for Study 204, the current M.O. found that abdominal pain and hypotension were more common in the cefepime arm.

Laboratory adverse events: Laboratory abnormalities in subjects with normal baseline values were infrequent with the exception of electrolyte imbalance. Abnormalities of renal function occurred similarly between the two treatment groups and no subject with a normal baseline creatinine developed a clinically relevant degree of renal insufficiency as determined by serum creatinine. Transaminase elevations were more frequent in the ceftazidime arm than the cefepime arm. No clinically relevant transaminase abnormalities developed in those receiving cefepime. Elevations of total bilirubin occurred in 12% of individuals receiving cefepime compared to 4% receiving ceftazidime but both groups had two subjects develop clinically relevant elevations. Electrolyte disturbances were common but similar between the two treatment arms. Hypophosphatemia reached clinically relevant ranges in 11% of subjects in both treatment arms.

The following was obtained from pages 50-51 of the JCSR and were the reviewers’ final comments/conclusions for Study AI411-204.

Final comments/conclusions - study 411-204 This was a large, multi-center, double-blind, randomized controlled trial comparing the efficacy of cefepime with that of ceftazidime for empiric therapy of febrile neutropenia. Noteworthy aspects of this trial include its reliance on the IDSA guidelines for key aspects of design. The trial enrolled a total of 276 patients, accounting for 315 episodes. Baseline demographic and prognostic factors were balanced between the treatment arms. Patients had a broad range of underlying diseases; although patients had been stratified by malignancy, comparability between treatment arms for patients with a given type of malignancy was not demonstrated. One hundred ninety-three (193) (61.3%) of enrolled patient episodes were found to be evaluable for efficacy by the FDA Medical Officer. The most common reasons for unevaluability were modification of the initial

270

regimen before assessment at 72 hours, fever due to a non-bacterial infection, absence of neutropenia, and loss to follow-up. The FDA analysis led to a larger number of patients being deemed unevaluable than in the sponsor’s analysis; this accounted for most of the differences in the efficacy analysis between the FDA and the sponsor. Efficacy rates in the evaluable population, as determined by the Medical Officer and assessed either in terms of resolution of the initial episode or survival of infection, were similar for cefepime and ceftazidime. This was true for all febrile episodes, first episodes, and microbiologically documented infections. The point estimate for the difference in response rates was -0.069 for all episodes and -0.061 for first episodes. In order to be deemed therapeutically equivalent as per the DAIDP Points to Consider document, the 95% confidence interval of the difference in cure rates between the test product and the control should lie above -0.20 and include zero. Based on these criteria, cefepime fails to establish therapeutic equivalence in the patient population deemed evaluable by the FDA reviewing Medical Officer when either the first or all febrile episodes are considered, as well as in patients with microbiologically documented infections. Thus, this study alone cannot demonstrate therapeutic equivalence between cefepime and ceftazidime. However, its design allows pooling of results with similar trials (see the Integrated Summary of Efficacy, section 10). Equivalence was shown under the MITT analysis. Safety analysis showed a higher all-cause mortality rate in the cefepime arm, primarily due to a higher incidence of secondary infection deaths in the cefepime arm. These deaths occurred after prolonged neutropenia, which is an independent risk factor for infection, and were due to a variety of pathogens. Of note, the incidence of treatment failures due to secondary infection in the FDA evaluable patient population was similar between treatment groups. Thus, there does not appear to be a clear association between cefepime administration and mortality due to new infections. Adverse event rates with cefepime were somewhat higher than in premarketing experience with this drug, presumably because of the use of a higher daily dose (2 g IV q8h). There was no significant difference in the incidence of clinical adverse events or the incidence of discontinuation due to clinical adverse events between treatment arms. There was no significant difference in the incidence of laboratory adverse events. In conclusion, study AI411-204 does not by itself demonstrate therapeutic equivalence between cefepime and ceftazidime for empiric therapy of febrile neutropenia; its design features allow pooling of its results with those of similarly designed and conducted studies. This study does demonstrate an acceptable safety profile for cefepime in this indication.

M.O. comment: The M.O notes that the JCSR found that “…comparability between treatment arms for patients with a given type of malignancy was not demonstrated”. The JCSR also found that the higher all-cause mortality rate in the cefepime arm was primarily due to a higher incidence of secondary infection deaths that occurred after prolonged neutropenia.

Study 189: General information from JCSR Title: A Multicenter Comparative Study of Cefepime vs. Ceftazidime as Empiric Therapy in the Treatment of Febrile Episodes in Neutropenic Patients. Objective: To evaluate the clinical efficacy and safety of cefepime administered at a dose of 2 grams every eight hours in comparison to ceftazidime administered at a dose of 2 grams every eight hours as empiric treatment of febrile episodes in neutropenic subjects ≥ 16 years of age. Study design: This was a multi-center, open-label, comparative, randomized (1:1) clinical trial conducted in Europe. A sample size of 120 subjects (60 per treatment arm) was originally planned as sufficient to rule out a difference in clinical response rate of >21% between treatment groups with 80% power. During the course of the study, it was decided to exceed this target of 120 in order to provide for more meaningful statistical comparisons and to extend the experience with cefepime in this indication. Study period: First subject enrolled February 20, 1993. Last subject completed therapy June 28, 1995. Two-hundred eighty-one subjects were enrolled and treated across 15 sites in northern Europe.

The JCSR M.O. noted the following related to the Study 189’s protocol. According to the study report, the original protocol was amended five times as follows. Amendment #1, which applied to all study sites, clarified that no new antibacterial, antifungal or antiviral agent should be initiated during the first 72-96 hours of the study; prophylactic antifungal or antiviral agents which were in place at the start of study therapy

271

could be continued. Amendment #2, which applied only to sites in the United Kingdom, made minor revisions to storage and mixing requirements consistent with that country’s regulations. Amendment #3, which applied to sites in the Netherlands, allowed for the concomitant use of prophylactic doses of certain quinolone antibiotics for suppression of gram negative bowel flora in subjects with hematologic malignancies. Amendment #4, which applied to Site #9 in Switzerland, specified circumstances for the modification of empiric therapy within first 72 hours. These included: the addition of vancomycin in the face of organisms resistant to the study drug, (e.g., methicillin-resistant Staphylococcus aureus (MRSA), or Corynebacterium jeikeium); or in the face of frank clinical deterioration in a subject with infection due to a gram positive organism. In addition, amikacin was to be added in this time interval in the face of microbiologically documented gram negative sepsis in profoundly neutropenic subjects (<100 PMN/µL). Amendment #5, which applied to sites in Sweden, provided for the exclusion of subjects with solid tumor only, revised slightly the procedures for the collection of bacterial cultures, and allowed for addition of an aminoglycoside or vancomycin in the first 72 hours of study therapy for specific clinical indications. An aminoglycoside was to be added to the empiric regimen for frank clinical deterioration or the persistence of bacteremia; vancomycin was to be added in the face of MRSA or C. jeikeium. Such additions were to constitute failures of the empiric regimen. Given that this was a multi-center trial, it is to be expected that different institutions should have varying practice patterns; it is, after all, one of the goals of multi-center trials to test therapies in different practice settings that approximate the real world. However, it is concerning that the protocol was amended so as to create a situation in which there would be non-uniform practice patterns. This raises the issue as to whether the need for these amendments was driven by a non-random distribution of patients at the involved centers, or whether these amendments, by not being applied to all centers, could lead to a bias in the outcome.

M.O. comment: The M.O. notes that this was an open-label study.

The JCSR noted concerns that the five protocol amendments were driven by a non- randomized distribution of patients at the involved centers. It may have also been the case that since these amendments were not applied to all centers, this may have led to bias in outcome. On page 54, the JCSR also noted that randomization was not stratified by underlying disease (hematologic malignancy or solid tumor). On page 56, the JCSR M.O. noted that many subjects’ data were only supplied for the end of therapy evaluation, without any further follow-up. This prevented assessment of the possibility of relapse, and this was true for both patients who were discontinued from study therapy and those who remained on study therapy for the time specified by the protocol.

Relevant Results from Study 189 Thirty-eight of the 281 subjects were randomized more than once for separate febrile episodes. In Study 189, the proportions of leukemia patients were similar between treatment groups 49.6% (69/139) for cefepime and 48.6% (69/142) for ceftazidime. Concomitant use of antibacterial prophylaxis was more common among ceftazidime subjects than among cefepime subjects (35% vs. 32%). During the first episode of febrile neutropenia, the proportions of deaths were equal between the two groups [6% (8/139) for cefepime and 6% (9/142) for ceftazidime]. The JCSR noted that as compared with Study 204, Study 189 had a significantly greater proportion of patients on prophylactic antimicrobial agents at study entry, as well as, during the study.

M.O. comment: The MO notes that both the proportions of acute leukemia patients and the number of deaths were similar between the two treatment groups

Study 131: General Information from JCSR Title. A Comparative Study of Cefepime vs. Ceftazidime in the Treatment of Adult Cancer Patients with Fever and Neutropenia.

272

Objective. To evaluate the clinical efficacy and safety of cefepime, administered at a dose of 2 grams every eight hours, in comparison to ceftazidime administered at a dose of 2 grams every eight hours for the empiric treatment of febrile episodes in neutropenic cancer subjects. Study design. A two arm, comparative, open-label, randomized (1:1) study conducted in the United States. The original protocol was amended to add a second site for adult subjects and a third site for enrollment of pediatric subjects. Ninety subjects were treated for a total of 104 febrile episodes at the two study sites. Initially, a single study site (Florida) planned to enroll 100 adult subjects over a period of approximately two years. A second adult study site (Missouri) was added and planned to accrue 100 adult subjects. Study site 001 (Florida) enrolled 87 subjects. One subject did not receive study drug, ten subjects were enrolled for two febrile episodes, and two subjects were enrolled for three. Study site 002 (Missouri) had difficulty enrolling subjects and the study was terminated early after four subjects were accrued. Study site 003 (Dallas) accrued 104 pediatric patients Study period for sites 1 and 2. First subject enrolled August 30, 1989. Last subject completed therapy November 26, 1991. Ninety subjects treated at 2 sites in the U.S. A third U.S. site was added later and accrued 104 pediatric patients. These patients were submitted separately as an efficacy supplement for the use of cefepime for treatment of pediatric infections. M.O. comment: The JCSR only analyzed the data from the adult sites.

The JCSR noted the following. The definition of neutropenia for this study was less stringent than in the other monotherapy studies (AI411-204 and 189), using a threshold of 1000 neutrophils/µL. In order to maintain consistency with analysis of the other studies, patients with an absolute neutrophil count >500 cells/µL were excluded from analysis by the Medical Officer.

The JCSR also noted that, “As with study AI411-189, patients were not stratified by underlying disease.”

Relevant Results from Study 131 Leukemia was the underlying disease in 60% (27/45) of ceftazidime patients and 55.6% (25/45) of cefepime patients. Duration of neutropenia (ANC< 500) > 7 days was 84.5% (38/45) for cefepime and 82.2% (37/45) for ceftazidime.

The JCSR noted the following: The proportion of patients with severe (<100 PMN/µL) or prolonged (duration of at least 1 week) neutropenia was greater in this study than in studies AI411-204 or 189. This may reflect the limited use of colony stimulating factors in patients in this study; it is worth noting that this study was conducted between 1989 and 1991, when use of such factors was not as common.

M.O. comment: The numbers of leukemia patients and durations of neutropenia were similar between the two groups. Any differences in proportions were likely magnified by the relatively small sample sizes in the two treatment arms.

The JCSR also noted the following. The overall response rate results for all episodes demonstrate equivalence between cefepime and ceftazidime, although the number of evaluable episodes is significantly less than in studies AI411-204 and -189, and the response rates in both arms of this study are lower than in studies AI411-204 and -189. These lower response rates may be have be due to the greater proportion of patients in this study with severe or prolonged neutropenia.

Regarding the safety analysis for Study 131, there were 10 deaths, 6 in the cefepime arm and 4 in the ceftazidime arm. The following table was obtained from page 87 of the JCSR.

273

Table 131.8 -Principal Cause of Death Number of Subjects Cefepime Ceftazidime Total (N=90) (N=45) (N=45) All deaths 6 (13.3%) 4 (8.9%) 10 (11.1%) Infection 2 (4.4%) 3 (6.6%) 5 (5.6%) Hemorrhage 2 (4.4%) 1 (2.2%) 3 (3.3%) Progression of Cancer 2 (4.4%) 0 (0.0%) 2 (2.2%) CMH p-value = 0.158

M.O. comment: Overall, the two treatment arms were balanced related to cause of death. Of note, the JCSR deemed that 2 of the cefepime deaths and none of the ceftazidime deaths were due to cancer progression.

The JCSR concluded that Study 131 supported “therapeutic equivalence between cefepime and ceftazidime for empiric therapy of febrile neutropenia, although the number of evaluable episodes studied was small.” Additionally, the JCSR noted the following. Pooling of this study with AI411-189 and -204 is problematic because of the difference in response rates, which, as described above, most likely reflects differences in study design and conduct. At the March 5, 1997 meeting of the Anti-Infective Drug Products Advisory Committee, the statistical consultant to the committee, Dr. Donald Parker stated that this trial should not be pooled with the other two monotherapy trials.

Study 118: General Information from JCSR Title: “A Multi-Investigator Comparative Study of Cefepime and Piperacillin/Gentamicin in the Treatment of Cancer Patients with Fever and Neutropenia.” Objective: To evaluate the clinical efficacy and safety of cefepime, administered at a dose of 2 grams every eight hours, in comparison to the combination of piperacillin administered at a dose of 3 grams every 4 hours and gentamicin administered at a dose of 1.5 mg/kg every 8 hours, for the empiric treatment of febrile episodes in neutropenic cancer subjects. Study design: A two arm, comparative, open-label, randomized (1:1) multi-center study conducted in the United States. A total of 116 subjects were treated at the two study sites. Study period: First subject enrolled June 23, 1989. Last subject completed therapy December 12, 1991.

Relevant Results from Study 118: Leukemia was the underlying disease in 33.9% (20/59) of cefepime patients and 28.1% (16/57) of comparator (control) patients. A solid tumor was the underlying disease in 28.8% (17/59) of cefepime patients and 40.3% (23/57) of control patients. The proportions of patients with an ANC nadir of < 100 were 37.3% (22/57) for cefepime and 54.4% (21/57) for control. Duration of neutropenia (ANC< 500) > 7 days was 52.5% (31/59) for cefepime and 49.1% (28/57) for control. According to the JCSR, 9 deaths (5 in the cefepime group and 4 in the control group) occurred within 30 days of termination of study therapy. None were considered related to study therapy. Two subjects in each group died of infection with either bacteremia or fungemia. The other five subjects died of complications associated with their underlying cancer or its treatment.

M.O. comment: There were slightly more leukemic patients in the cefepime arm, and more solid tumor patients in the comparator arm. More ceftazidime patients had an ANC nadir of < 100; however, the patients in both groups had similar durations of neutropenia. This may

274 have been related to the fact that colony stimulating factors were used more frequently in the control group (11%) as compared with the cefepime group (3%). The total number of deaths within 30 days of study therapy was similar in the two groups.

Study 137: General Information from JCSR Title: A Comparative Study of Cefepime vs. Mezlocillin/Gentamicin in the Treatment of Fever in Cancer Subjects with Fever and Neutropenia. Objective: To evaluate the clinical efficacy and safety of cefepime administered at a dose of 2 grams every eight hours in comparison to the combination of mezlocillin administered at a dose of 3 grams every 4 hours and gentamicin administered at a dose of 1.5 mg/kg every eight hours, for the empiric treatment of febrile episodes in neutropenic cancer subjects. Study design: A two arm, comparative, open-label, randomized (1:1) single-center study conducted in the United States. A total of 200 subjects were planned with a projected accrual of 2 to 3 subjects per week over a period of two years. However, enrollment was terminated at the end of two years with 71 subjects enrolled. Study period: First subject enrolled July 28, 1990. Last subject completed therapy August 14, 1992.

Relevant Results from Study 137: Leukemia was the underlying disease in 11.4% (4/35) of cefepime patients and 16.7% (6/36) of comparator (control) patients. A solid tumor was the underlying disease in 11.4% (4/35) of cefepime patients and 11.1% (4/36) of control patients. The proportions of patients with an ANC nadir of < 100 were 97.1% (34/35) for cefepime and 100% (36/36) for controls. Duration of neutropenia (ANC< 500) > 7 days was 100% (35/35) for cefepime and 94.4% (34/36) for controls. Bone marrow transplant was performed in 97.1% (34/35) cefepime patients and 94.4% (34/36) controls. According to the JCSR, 2 deaths (1 in the cefepime group and 1 in the control group) occurred within 30 days of termination of study therapy. None were considered related to study therapy. The cefepime patient died due to “alveolar hemorrhage syndrome” and the control patient died due to veno-occlusive disease. The other five subjects died of complications associated with their underlying cancer or its treatment.

M.O. comment: The treatment arms appeared to be balanced on baseline demographic factors. The total number of deaths within 30 days of study therapy was equal in the two groups.

Study 186: General Information from JCSR Title: A Multi-Investigator Comparative Study of Cefepime and Ceftazidime, in Combination with Amikacin in the Treatment of Patients with Fever and Neutropenia. Objective: To evaluate the safety and efficacy of cefepime (administered at 2 g q12h) and amikacin (administered at 7.5 mg/kg q12h) versus ceftazidime (administered at 2 g q8h) and amikacin in the treatment of subjects with fever and neutropenia. Study design: A two arm, comparative, open-label, randomized (2:1) multi-center study conducted in France. Enrollment of approximately 300 subjects was planned. The actual accrual was 353 subjects at 31 sites. Each subject was treated for a single febrile episode during a period of neutropenia. Study period: First subject enrolled October 10, 1992. Last subject completed therapy November 4, 1993.

275

M.O. comment: An amendment to the protocol modified the randomization schedule from 1:1 to 2:1 (cefepime: comparator) and increased the sample size from 200 to 300+.

On page 109 of the JCSR, the M.O. stated the following regarding the cefepime study therapy dose of 2 g IV q12 hrs in combination with amikacin. The dosage for cefepime in this study was 50% lower than in the other trials. The rationale for choosing this dose was not explicitly stated in the protocol. This discrepancy between this dosage and the dosage proposed for this indication raises questions about extrapolation of safety data from this application to situations in which cefepime is used at 2 g IV q8h in combination with an aminoglycoside. In addition, comparison of efficacy rates in this study with those of other studies is problematic; one might expect lower efficacy rates, particularly for Gram-positive organisms, given that the clinical course of infected neutropenic patients depends strongly on the use of appropriate doses of antibiotics.

M.O. comment: The JCSR M.O. had concerns that the lower cefepime dose would result in lower efficacy rates for cefepime.

Relevant Results from Study 186: Leukemia was the underlying disease in 58.7% (142/242) of cefepime patients and 58.6% (65/111) of comparator (control) patients. A solid tumor was the underlying disease in 4.1% (10/242) of cefepime patients and 4.5% (5/111) of control patients. The proportions of patients with an ANC nadir of < 100 were 83.5% (202/242) for cefepime and 86.5% (96/111) for controls. Duration of neutropenia (ANC< 500) > 7 days was 84.3% (204/242) for cefepime and 83.8% (93/111) for controls. Bone marrow transplant was performed in 40.1% (97/242) cefepime patients and 42.3% (47/111) controls. The JCSR reported that the use of colony stimulating growth factors was similar between the treatment groups.

On page 119, the JCSR M.O. noted the following related to the antimicrobial prophylaxis practices employed in this study. The prophylaxis practices in this study make interpretation of any response rates problematic. The efficacy of prophylaxis against infection in this setting has not been demonstrated, and the use of parenteral antibiotics as prophylaxis, especially when continued during therapy, confuses interpretation of efficacy. Use of this sort of prophylaxis prevents attribution of a therapeutic effect to the study regimen alone, thereby preventing strong conclusions as to the efficacy of the study regimen. Furthermore, from a regulatory perspective, it is not clear how drugs studied under these conditions should be labeled, except perhaps with multiple qualifiers as to their use (e.g., ‘efficacy shown only in the setting of prophylaxis with the following regimens’). The IDSA guidelines recommend that if patients on prophylaxis are enrolled into a study of empiric therapy of febrile neutropenia, then all patients should be on the same, identifiable regimen. Failing that, such patients should be stratified or excluded. Since a wide variety of prophylactic regimens were used, without stratification, the analysis performed by the Medical Officer excluded these patients.

According to the JCSR, 17 deaths [13/242 (5.4%) in the cefepime group and 4/111 (3.6%) in the control group] occurred within 30 days of termination of study therapy. None were considered related to study therapy.

M.O. comment: The treatment arms appeared to be balanced on baseline demographic factors. The proportions of deaths within 30 days of study therapy were similar in the two groups.

276

Study 198: General Information from JCSR Title: A Multicenter, Randomized, Prospective, Comparative Study of Cefepime Plus Vancomycin Versus Ceftazidime Plus Vancomycin as Empiric Therapy in the Treatment of Febrile Episodes in Granulocytopenic Patients With Hematological Malignancies Objectives: To evaluate the safety and clinical efficacy of cefepime, 2 g q8h, plus vancomycin, 30 mg/kg/day compared with ceftazidime, 2 g q8h, plus vancomycin, 30 mg/kg/day, as empiric therapy in the treatment of febrile neutropenic episodes. Study design: A multicenter, open, randomized (1:1), prospective clinical trial conducted in Belgium. One hundred and eleven subjects were treated for a total of 128 febrile episodes at the four study sites. There were 53 subjects in the cefepime/vancomycin group and 58 in the ceftazidime/vancomycin group. Study period: First subject enrolled 3 February 1993; last subject completed therapy 21 February 1994.

Relevant Results from Study 198: Leukemia was the underlying disease in 59.3% (32/54) of cefepime patients and 52.6% (30/57) of comparator (control) patients. A solid tumor was the underlying disease in 3.7% (2/54) of cefepime patients and 1.8% (1/57) of control patients. The proportions of patients with an ANC nadir of < 100 were 92.6% (50/54) for cefepime and 93% (53/57) for controls. Duration of neutropenia (ANC< 500) > 7 days was 79.7% (43/54) for cefepime and 73.7% (42/57) for controls. Bone marrow transplant was performed in 24.1% (13/54) cefepime patients and 21.1% (12/57) controls. Antimicrobial prophylaxis was used in 87% (47/54) cefepime patients and 91.2% (52/57) control patients. The JCSR reported that the use of colony stimulating growth factors was similar between the treatment groups.

According to the JCSR, 8 deaths [5 in the cefepime group and 3 in the control group] occurred within 30 days of termination of study therapy. One of the control patients’ deaths was thought to be related to study therapy. (The patient developed renal insufficiency that was attributed to vancomycin.)

M.O. comment: The treatment arms appeared to be balanced on baseline demographic factors. The numbers of deaths within 30 days of study therapy were similar in the two groups (5/54 in the cefepime group and 3/57 in the control group); however, due to the small sample sizes, the percentage of patients that died in the cefepime arm (9%) was nearly twice that in the comparator arm (5%).

TWO PHASE 2 NON-COMPARATIVE STUDIES Study 143: General Information from JCSR Title: Non-Comparative Study of Cefepime as Empiric Therapy of Fever in Cancer Patients with Granulocytopenia Objective: This study was intended to provide “additional safety data and preliminary efficacy data for cefepime administered intravenously in the empiric treatment of cancer patients with fever and neutropenia”. Design: Open-label, non-comparative Phase II trial. 84 patients representing 108 episodes were enrolled in Belgium and Switzerland. Study period: First patient enrolled January 28, 1990; last patient assessed June 25, 1991

Relevant Results from Study 143:

277

Leukemia was the underlying disease in 39% of the patients. A solid tumor was the underlying disease in 34% of the patients. The JCSR M.O. stated on page 138 that, “No data were provided on the duration or severity of neutropenia in this population”.

According to the JCSR, there were 10 deaths: 5 during treatment, 4 within 30 days post therapy, and one 39 days post therapy. None of the deaths were attributed to cefepime use. Four of the deaths were reportedly associated with the infection under treatment. In 3 patients, a new infection contributed to the deaths. Underlying disease was reported as a contributing factor for all of the deaths.

M.O. comment: This was a Phase 2 study, and rates of bone marrow transplant, antimicrobial prophylaxis, and use of colony stimulating factors were not discussed. The 30-day mortality rate was 10.7% (9/84).

Study 158: General Information from JCSR Title: A Noncomparative Prospective Multicenter Study Evaluating Efficacy and Safety of Cefepime as Empirical Therapy of Febrile Episodes in Granulocytopenic Patients Objectives: To evaluate the safety and efficacy of cefepime as monotherapy in the empiric treatment of fever in neutropenic patients; and to evaluate the outcome of cefepime treatment with or without treatment modification in neutropenia patients presenting with fever. Design: This was a single center, open-label, non-comparative pilot trial. A total of 30 patients were enrolled in the Netherlands. Study Period: First subject enrolled 2/3/91. Last subject completed therapy 11/20/91.

Relevant Results from Study 158: Leukemia was the underlying disease in 77% of the patients. Bone marrow transplant occurred in 96.7% of patients. Antimicrobial prophylaxis was given to 87% of patients. Severe neutropenia (ANC<100) was observed in 80% of patients. There were 4/30 (13.3%) deaths within 30 days post therapy and 1 death reported at 54 days post therapy. None were considered related to cefepime.

M.O. comment: This was a Phase 2 study. Death rates in both noncomparative Phase 2 studies were higher than those observed in the comparative Phase 3 studies.

JCSR Integrated Analysis of Efficacy for Febrile Neutropenia: The following was obtained from the Integrated Analysis of Efficacy found on pages 145-155 of the JCSR. Pooling of trials Two of the three monotherapy trials with ceftazidime as the active control (AI411-204 and 189) had similar designs, were suitable for consistent application of a uniform set of evaluability criteria, had individual response rates which did not show unusual variability, and had comparable individual attrition rates. Thus, pooling of these studies was justifiable. The third monotherapy study, AI411-131, although sharing design features with AI411-204 and 189, differed sufficiently that pooling of this study was not felt to be justified. Study AI411-131 differed in its design (essentially a single center trial), inclusion criteria (using a definition of neutropenia of <1000 cells/µL), exclusion criteria (excluding significant classes of infected patients included in studies 189 and 204), patient demographics (a higher proportion of patients had severe or prolonged neutropenia) and conduct (a considerably lower proportion of patients in study 131 received colony stimulating factor support). Finally, study 131 was conducted considerably earlier than the other monotherapy studies and showed significantly lower response rates, which may have been related

278

to the proportion of patients with severe or prolonged neutropenia. During the March 5, 1997 meeting of the Anti- Infective Drug Products Advisory Committee, the statistical consultant to the committee, Dr. Donald Parker stated that he would recommend against pooling study 131 with the other monotherapy trials.

M.O. comment: Based on the JCSR comments, Study 131 was not pooled with Studies 204 and 189 because of several reasons: single center study, different inclusion and exclusion criteria, higher proportion of patients with prolonged neutropenia and less use of G-CSF, the study was conducted much earlier than the other monotherapy studies, and had significantly lower response rates.

When meta-analytic methods for pooling (DerSimonian and Laird 1986) were applied to studies AI411-204 and 411 189, the p-value obtained from the Breslow-Day test for heterogeneity of trials was 0.9427 (a p-value <0.05 was considered statistically significant). Therefore, these two monotherapy studies showed homogenous treatment effects and were pooled; unless otherwise indicated, pooled results from studies AI411-189 and AI411-204 will be presented below in subsequent analyses. The combined response rates for the pooled studies in the evaluable population were 95/192 (49.5%) for cefepime and 98/176 (55.7%) for ceftazidime, with a point estimate for the difference in response rates of -6.2%. The 95% confidence interval obtained by the DerSimonian-Laird pooling procedure was 192, 176 (-0.1663, 0.0356) 49 5%, 55 7% for the two pooled monotherapy studies, using definition 1B. This indicates that cefepime is therapeutically equivalent to ceftazidime in the pooled population database. Some patients were enrolled multiple times in these trials. Because episodes in the same patient may not represent independent events, analysis of first episodes in the pooled patient database was performed, as recommended in the IDSA guidelines. This analysis gave response rates of 83/164 (50.6%) for cefepime and 84/153 (54.9%) for ceftazidime in the pooled evaluable patient population, with a point estimate for the difference in response rates of 4.3%. The 95% confidence interval around the difference in response rates was 164,153(-0.1591, 0.0733)50 6%,54 9%. This indicates that cefepime is therapeutically equivalent to ceftazidime in the pooled population database when analyzed by patient. The two studies that had combination therapies as their active control (Study Group 2; studies AI411-118 and 137) were deemed not to be poolable on the basis of use of different comparators, different patient populations, and significantly different response rates. The p-value obtained from the Breslow-Day test was 0.0152 for these two studies. This statistically significant result indicates that there may be trial by treatment interaction in studies AI411 118 and AI411-137, precluding pooling. Confidence intervals for differences in response rates (for all episodes and first episodes) for individual studies, as well as the pooled monotherapy studies are shown in Figures 1A and B. Reasons for failure A summary of reasons for failure in the pooled evaluable database is shown in Table 10.1. Table 10.1. Reasons for treatment failure - pooled evaluable database Reason for failure Cefepime Ceftazidime (N=192) (N=176) Persistent fever 37 (19.3%) 27 (15.3%) Poor microbiologic response 11 (5.7%) 11 (6.2%) - initial isolate resistant Poor microbiologic response 3 (1.6%) 1 (0.6%) - initial isolate susceptible Death from primary infection 4 (2.1%) 7 (4.0%) Death from secondary 5 (2.6%) 0 infection Poor clinical response 12 (6.2%) 10 (5.7%) Bacteriologic relapse 0 0 New MDI - susceptible 0 0 isolate

279

New MDI - resistant isolate 6 (3.1%) 9 (5.1%) New CDI 1 (0.5%) 4 (2.3%) New FUO 18 (9.4%) 9 (5.1%) Total failures 97 (50.5%) 78 (44.3%)

Except for deaths due to secondary infection, there was not a significant difference between the treatment arms with respect to reasons for treatment failure. The deaths due to secondary infection in the cefepime arm were due to enterococcal bacteremia in two cases, an aspiration pneumonia in one case, and Aspergillus pneumonia in the fourth; the other death was attributed to infection but occurred in the setting of fever without a documented source of infection. Efficacy with respect to different definitions of outcome As discussed in the reviews for the individual studies, the analysis of treatment failures indicated a broad distribution of reasons for discontinuation or modification of the empiric regimen. Some of these outcomes may have masked successful aspects of treatment (e.g., resolution of the initial episode with failure to prevent further infections while on therapy). Response rates were therefore analyzed using different measures of outcome, as described in Methods. Use of this method was endorsed by the Advisory Committee at the March 5, 1997 AIDPAC meeting.

M.O. comment: The first analysis, as demonstrated in Figure 1A, included all episodes of febrile neutropenia. (Patients were allowed to enroll more than once if they had additional episodes of febrile neutropenia.) The second analysis, Figure 1B, included only the first episode of febrile neutropenia.

280

Figure 1A. 95% confidence intervals around differences in response rates analyzed by episode (i.e., all episodes). Heavy vertical bars indicate point estimates of differences in response rates between cefepime and comparator. The dashed lines indicate the therapeutic equivalence zone. The negative side of axis favors comparator; positive side favors cefepime.

-50 -20 0 20 50 Study

AI411-118 (cefepime vs. piperacillin/ gentamicin)

AI411-137 (cefepime vs. mezlocillin/ gentamicin)

AI411-131 (cefepime vs. ceftazidime)

AI411-204 (cefepime vs. ceftazidime)

AI411-189 (cefepime vs. ceftazidime)

AI411-189 + AI411-204

281

Figure 1B. 95% confidence intervals around differences in response rates analyzed by patient (i.e., first episodes only). Heavy vertical bars indicate point estimates of differences in response rates between cefepime and comparator. The dashed lines indicate the therapeutic equivalence zone. The negative side of axis favors comparator; the positive side favors cefepime.

-50 -20 0 20 50 Study

AI411-118 (cefepime vs. piperacillin/ gentamicin)

AI411-137 (cefepime vs. mezlocillin/ gentamicin)

AI411-131 (cefepime vs. ceftazidime)

AI411-204 (cefepime vs. ceftazidime)

AI411-189 (cefepime vs. ceftazidime)

AI411-189 + AI411-204

282

Efficacy in clinically relevant subpopulations The Medical Officer identified a number of patient subpopulations of special clinical interest. It must be noted that the studies were neither designed nor powered to test hypotheses associated with these subpopulations defined post hoc. Furthermore, these subsets are not mutually exclusive nor has any adjustment in the test level been taken to account for this or the multiple comparisons bias. As noted in the previous section, these statistical results need to be interpreted with care because of the possibility that multiple post hoc comparisons of the data may lead to Type I errors. These comparisons are offered only as a means of ascertaining homogeneity of conclusions regardless of underlying condition(s) of patients treated. Any highly significant differences uncovered will only generate a hypothesis to be further explored for clinical importance. Table 10.3 summarizes the efficacy of cefepime in subpopulations identified as clinically important by the Medical Officer. These analyses are based on patients deemed evaluable by the Medical Officer in the pooled monotherapy database with ceftazidime as the active control. Definition 1B was taken to be the criterion for success. No significant difference in response rates was established between cefepime and comparator for patients with major risk factors for infection (severe or prolonged neutropenia, leukemia as the underlying disease, history of bone marrow transplantation, and presence of an indwelling catheter. With regard to severity of infection at presentation, the only physiologic variable available for subset analysis was systolic blood pressure; the subset of patients who were hypotensive (SBP < 90 mm Hg) at presentation was too small to allow statistically valid comparisons. Therapeutic equivalence between the two treatment groups was not established in patients with hematologic malignancies other than leukemia (e.g., lymphoma), patients who did not have severe neutropenia (ANC>100), and when response rates were analyzed by individual infection type. In interpreting these results, it should be emphasized that these studies were not intended or designed to demonstrate therapeutic equivalence between the treatment arms for these subpopulations. Although the response rates are comparable between the treatment arms for these subgroups, no conclusions can be drawn with regard to therapeutic equivalence for specific subgroups, since these represent post hoc analyses, and comparability of the treatment groups when retrospectively stratified by these subgroups cannot be assured. It is, however, interesting, that patients with factors thought to represent an increased risk for infection (severe or prolonged neutropenia, presence of an indwelling catheter, a hematologic malignancy, bone marrow transplant) had lower response rates in both arms than patients without such factors. M.O. comment: As was reflected in the labeling, cefepime monotherapy is not indicated in the treatment of febrile neutropenic patients with severe or prolonged neutropenia, the presence of an indwelling catheter, a hematologic malignancy, or bone marrow transplant. The JCSR noted that in the pooled analysis response rates for such patients were lower in both study arms.

283

Table 10.3. Response Rates in clinically relevant subpopulations1 Subpopulation Cefepime Ceftazidime 95% Confidence Interval

First febrile episode only 83/164 84/153 164, 153 (-0.1591, 0.0733) 50 6%, 54 9% (50.6%) (54.9%)

•Patients with leukemia 30/82 27/64 82, 64 (-0.2296, 0.1176) 36 6%, 42 2% (36.6%) (42.2%)

•Patients with other hematologic malignancies 19/50 21/41 50, 41 (-0.3581, 0.0937) 38 0%, 51 2% (38.0%) (51.2%) Patients with other hematologic diseases 1/2 (50.0%) 3/9 (33.3%) Insufficient sample size

Patients with solid tumors 45/58 47/62 58, 62 (-0.1502, 0.1858) 77 6%, 75 8% (77.6%) (75.8%)

•Patients on prophylactic antibiotics 28/75 28/66 75, 66 (-0.2270 , 0.1252) 37 3%, 42 4% (37.3%) (42.4%)

Patients not on prophylactic antibiotics 67/117 70/110 117, 110 (-0.1995, 0.0721) 57 3%, 63 6% (57.3%) (63.6%)

Patients with severe neutropenia (ANC ≤ 100) 65/134 67/132 134, 132 (-0.1502, 0.1052) 48 5%, 50 8% (48.5%) (50.8%)

•Patients with ANC > 100 30/58 31/44 58, 44 (-0.3936, 0.0190) 51 7%, 70 5% (51.7%) (70.5%)

•Patients with MDIs 24/62 23/54 62, 54 (-0.2353, 0.1576) 38 7%, 42 6% (38.7%) (42.6%)

•Patients with CDIs 12/26 11/19 26, 19 (-0.4562, 0.2214) 46 2%, 57 9% (46.2%) (57.9%)

•Patients with FUOs 59/104 64/103 104, 103 (-0.1973, 0.0892) 56 7%, 62 1% (56.7%) (62.1%)

Patients without prolonged neutropenia (< 7 d) 64/100 70/105 100, 105 (-0.1667, 0.1134) 64%, 66 6% (64%) (66.7%)

•Patients with prolonged neutropenia (≥ 7 d) 31/92 28/71 92, 71 (-0.2191, 0.1042) 33 7%, 39 4% (33.7%) (39.4%)

•Patients with an indwelling catheter 42/112 48/103 112, 103 (-0.2319, 0.0499) 37 5%, 46 6% (37.5%) (46.6%)

Patients with no indwelling catheter 53/80 50/73 80, 83 (-0.1842, 0.1393) 66 3%, 68 5% (66.3%) (68.5%)

•Bone marrow transplant recipients 1/10 (10%) 1/7 (14.3%) 10, 7 (-0.4833, 0.3976) 10%, 14 3%

1 The 95% confidence intervals are reported as nt,nc ( 95% C.I.) pt,pc where nt = number in the test group, nc = number in the control group, pt = response rate in the test group, pc = response rate in the control group. • denotes lack of therapeutic equivalency between cefepime and ceftazidime in the subpopulation.

284

Table 10.3. Response Rates in clinically relevant subpopulations1 Subpopulation Cefepime Ceftazidime 95% Confidence Interval

No bone marrow transplant 94/182 97/169 182, 169 (-0.1673, 0.0523) 51 6%, 57 4% (51.6%) (57.4%) Patients with systolic BP <90 at entry 2/8 (25.0%) 1/2 (50.0%) Insufficient sample size

Patients with systolic BP ≥90 at entry 93/184 97/174 184, 174 (-0.1343, 0.0652) 50 5%, 55 7% (50.5%) (55.7%)

285

Efficacy based on the causative pathogen 62 patients in the cefepime arm and 54 patients in the ceftazidime arm in the evaluable pooled population had microbiologically documented infections. Response rates (using definition 1B) for the most common causative bacterial pathogens are shown in Table 10.4. It should be emphasized that the studies were neither designed nor powered to demonstrate a difference in treatment effect between the arms with regard to specific pathogens.

Table 10.4 Response by causative pathogen Pathogen Cefepime Ceftazidime p-value* Success Failure Total Success Failure Total E. coli 2 11 13 7 4 11 0.033 S. epidermidis 3 10 13 2 7 9 1.000 K. pneumoniae 2 3 5 3 4 7 1.000 P. aeruginosa 2 6 8 1 0 1 0.333 S. mitis 4 1 5 2 4 6 0.242 Enterococcus spp. 0 5 5 0 6 6 — S. viridans 0 0 0 0 0 0 — S. aureus 4 1 6 0 1 1 0.333 * By Fisher’s exact test (two-tailed) There was no statistically significant difference in response rates between the two treatment arms, except for Escherichia coli, which was the pathogen most frequently identified in MDIs. (It should be remembered that this is a post hoc analysis, with the potential for a type I error. The difference in response rates between treatment arms for E. coli infections may reflect noncomparability of the groups, rather than a true difference in response rates.) The most common reasons for treatment failure for E. coli infections in the cefepime arm were persistent fever (3 cases), a new microbiologic documented infection with a different organism (3 cases), and resistance of the original isolate (2 cases). There was one death due to primary infection with E. coli in the cefepime arm. Thus, the lack of equivalence for E. coli infections for cefepime reflected the occurrence of secondary infections which were not connected with the original infection. In other words, this outcome definition asks the drug to act not only as a therapeutic agent for the initial infection, but also as a prophylactic agent during therapy against a secondary infection, an event which is independent of the primary infection. If outcome measure 2 was used - that is, outcome of the primary infection without regard to subsequent infections - the response rates for E. coli infections were 8/11 (72.7%) for ceftazidime, and 7/13 (53.8%) for cefepime. The difference between response rates under outcome measure 2B was not statistically significant (p=0.422 by Fisher’s exact test).

M.O. comment: There was only one death in the cefepime arm due to a primary infection with E. coli. The JCSR noted that, “The difference in response rates between treatment arms for E. coli infections may reflect noncomparability of the groups, rather than a true difference in response rates.” The JCSR also noted that, “…the lack of equivalence for E. coli infections for cefepime reflected the occurrence of secondary infections which were not connected with the original infection.”

286

The following was the JCSR Final conclusions regarding efficacy found on pages 156-157. The applicant has submitted seven studies, five of them controlled, assessing the efficacy of cefepime as monotherapy for febrile neutropenia. The two largest studies, AI411-204 and 189, were designed according to the IDSA guidelines, and are adequate and well-controlled studies as defined in 21 CFR 314.126. The other three monotherapy studies (AI411-131, 118, and 137), although designed and conducted prior to publication of the IDSA guidelines, also meet the definition of adequate and well-controlled studies. Individually, results from AI411-204 and 189 do not show therapeutic equivalence between cefepime and the control regimen. However, the pooled results from studies AI411-204 and AI411-189, as assessed by a variety of outcome measures, and the results from study AI411-131, demonstrate therapeutic equivalence between cefepime monotherapy and ceftazidime monotherapy for empiric therapy of febrile neutropenia. This conclusion is further supported by the results from the studies comparing cefepime to combination therapy (AI411-118 and AI411-137), and the uncontrolled studies (AI411-143 and 158). Therefore, the sponsor has met the requirement stated in the action letter of July 26, 1994 for the original NDA (see section 3) of submitting results from two independent, adequate and well-controlled studies which demonstrate efficacy of cefepime for this indication. The combination therapy studies (AI411-186 and AI411-198) were marked by problems in their conduct resulting in a large number of patients being unevaluable for efficacy. The data from these studies does not have sufficient power to demonstrate therapeutic equivalence between cefepime in combination with vancomycin or an aminoglycoside, and the corresponding ceftazidime combination. Thus, these studies were not adequate to demonstrate the efficacy of cefepime in combination with an aminoglycoside or glycopeptide for empiric therapy of febrile neutropenic patients. In conclusion, the integrated analyses of efficacy and safety (see section 11) demonstrate that cefepime monotherapy is safe and effective for the empiric therapy of febrile episodes in neutropenic patients. The analyses do not demonstrate efficacy of cefepime in combination with an aminoglycoside or glycopeptide for this indication.

M.O. comment: The studies of cefepime in combination with an aminoglycoside or glycopeptide versus comparable combination therapy with ceftazidime were not sufficiently powered with evaluable patients to demonstrate efficacy. Therefore, the label communicated that cefepime used in combination with other antibacterial therapy was not indicated for empiric treatment of patients with febrile neutropenia.

JCSR Integrated Analysis of Safety for Febrile Neutropenia: The following was obtained from the JCSR Integrated Analysis of Safety found on pages 158 170. Mortality All cause mortality and deaths due to primary infection, secondary infections, underlying disease, and other causes were analyzed by treatment arm for all enrolled patients in Study Groups 1 and 2, as shown in Table 11.1. Table 11.1. Mortality in monotherapy studies Cause Cefepime Control (N=455) Overall (N=930) (N=475) All 40 (8.4%) 30 (6.4%) 70 (7.5%) 1° infection 10 (2.1%) 11 (2.4%) 21 (2.2%) 2° infection 12 (2.5%) 7 (1.5%) 19 (2.0%) Underlying 17 (3.6%) 11 (2.4%) 28 (3.0%) disease Other 1 (0.2%) 1 (0.2%) 2 (0.2%)

There did not appear to be a significant difference in mortality either overall or due to specific causes. Kaplan-Meier analysis did not reveal any significant differences between treatment arms with regard to time to death.

M.O. comment: As was noted in Table 11.1, there were slightly more cefepime deaths due to secondary infections and underlying disease. However, these differences did not reach statistical significance.

287

Adverse Events Very few adverse clinical events were reported in the 114 subjects accrued in the two non-comparative Phase II studies (AI411-143 and AI411-158; see Table 11.2 at the end of section 11). The most frequent were rash and diarrhea, which occurred in a total of 8 and 7 subjects, respectively. Erythema and nausea were reported in 3 subjects each. All the other adverse events occurred in single subjects; in most instances, they were considered to be not drug-related. The comparative Phase III studies of cefepime monotherapy provide the core of the analysis of adverse clinical events (see Table 11.2 at the end of the section 11). A graphical presentation of adverse event rates is shown in Figure 3. Despite differences between the individual studies, the patterns of adverse events observed in all five trials were consistent and comparable across treatment groups. Overall, the most frequent adverse event was diarrhea, which was observed in about one-fifth of the subjects. The highest incidence was reported in two studies: 40% in the cefepime arm of AI411-131 and 75% in the combination arm of AI411-137. The other frequent adverse events were rash and gastrointestinal disturbances such as nausea, abdominal pain, and vomiting. A number of general symptoms, such as headache, fever, and chills, were also commonly reported. Some important differences were noted across the studies. The highest incidence of adverse events was seen in study AI411-137, which included many subjects who had undergone bone marrow transplantation, while the lowest was reported in studies AI411-118 and AI411-204, which had the highest proportion of subjects with solid tumors and therefore the shortest duration of neutropenia. JCSR Medical Officer’s Comment Gastrointestinal disturbances are particularly common in patients undergoing treatment of malignancy, because of chemotherapy-induced destruction of gastrointestinal mucosa.

288

289

In the pooled analysis of the comparative Phase III studies, the incidence of adverse clinical events was very similar in the cefepime and ceftazidime groups. The most frequent was diarrhea, which was observed in 17% of the subjects in both groups. Only a limited number of these episodes were considered to be drug-related, with an overall incidence of 2% for cefepime and 1% for ceftazidime. While other gastrointestinal disturbances, such as abdominal pain and vomiting, were somewhat more frequent in the cefepime group, the majority of these events were either not drug- related or were of unknown relationship to therapy. The incidence of the other most frequent adverse events was similar in both treatment groups. There were very few probably drug-related adverse events. The comparison of cefepime to the two gentamicin-based combinations, AI411-118, and -137, yielded somewhat different results. There were differences detected between cefepime and the combinations for a number of adverse events, particularly gastrointestinal disturbances. The incidence of diarrhea was 41% in the combination group and 28% in the cefepime group. Nineteen percent of the diarrhea cases were felt to be probably drug-related, compared to 13% of the cases in the cefepime group. While the overall incidence of nausea was lower in the cefepime group, 20% compared to 25% in the combination group, the incidence of probably related nausea (9% versus 4%) was higher in the cefepime group, as was the incidence of vomiting (7% versus 2%). The pattern of other adverse events was similar between cefepime and the combinations and were similar to those events reported in the comparison of cefepime to ceftazidime. The most frequent adverse events occurred in the Phase III studies of cefepime in combination, AI411-186 and AI411 198. Both rash and diarrhea were less commonly reported in these studies than in those of cefepime monotherapy, while abnormal kidney function was more commonly reported. Of the 32 instances of abnormal kidney function in study AI411-186, only one, in the cefepime/amikacin group, was attributed to study therapy; four of the seven cases of abnormal kidney function that developed in study AI411-198 were attributed to study therapy, two in each treatment group. Mucositis and fever were the most frequently reported adverse events in studies AI411-186 and AI411-198, respectively. Probably Drug-related Adverse Events There was a total of 10 probably drug-related adverse events in the pooled analysis of the two non-comparative Phase II studies. They consisted primarily of rash (4 subjects) and diarrhea (3 subjects). Other probably drug-related adverse events were nausea, abdominal pain and erythema. These adverse events were generally mild; there were, however, 4 episodes of moderate rash. Probably drug-related adverse events were also uncommon in the randomized trials. The most frequent was rash. The frequency of this adverse event was similar in all five studies, with the exception of study AI411-137, in which about one-third of the subjects experienced probably drug-related rash. The majority of the other probably drug-related events occurred in single subjects. There were, however, some notable exceptions, all occurring in study AI411-137. Diarrhea was common; a larger proportion of subjects in the combination group experienced this side effect (47% combinations versus 31% cefepime). The other exception was the high incidence of nausea (20% for cefepime and 8% for the combination) and vomiting (17% and 6%, respectively). In the comparison of cefepime to ceftazidime, rash was the single most frequent probably drug-related adverse event. In most instances it was mild to moderate; in 3 subjects (2 cefepime and 1 ceftazidime) the rash was judged by the investigator to be severe. The incidence of other probably drug-related adverse events was similar between the cefepime and ceftazidime groups, with most rates being 1% or lower. Almost all events were mild to moderate, with the exception of 3 subjects who experienced severe adverse events other than rash. In the comparison of cefepime to combination therapy, the most frequent probably drug-related adverse event was diarrhea, which was seen in 13% of cefepime subjects and 19% of the combination group. Diarrhea was usually mild to moderate; one subject in the combination group had diarrhea that was considered severe. The incidence of probably drug-related rash was 13% for cefepime and 14% in the combination group. Once again, most events were mild to moderate; three subjects (2 cefepime and 1 combination) developed severe rashes. With the exception of nausea and vomiting, all other adverse events were usually reported in single subjects. Of note, there were 2 episodes of severe kidney failure in the combination group. These episodes were felt to be related to the aminoglycoside used in the combination regimens. Local Intolerance Of the 114 subjects enrolled in the non-comparative trials, AI411-143 and AI411-158, only one experienced local intolerance of cefepime, in the form of phlebitis. In the overall analysis of the randomized trials of cefepime monotherapy, the overall incidence of local intolerance of study therapy was 6% for cefepime subjects and 4% for control subjects. Phlebitis and infiltration of the IV catheter site were the most common manifestations of intolerance. In the comparison of cefepime to ceftazidime, local intolerance was greatest in study AI411-131 and somewhat higher in the cefepime arm (16% versus 11% in the ceftazidime arm). Local intolerance was extremely uncommon in the comparison of cefepime to combination therapy.

290

Changes in Laboratory Values Subjects with Normal Baseline Values Laboratory abnormalities were frequently encountered in the non-comparative studies. Of the assessed subjects with normal baseline values, one half developed hypocalcemia and hyperphosphatemia. Thirty-one subjects (47%) developed elevations in ALT, nine of which were clinically relevant. Other abnormalities were seen in about one- quarter of the subjects.

M.O. comment: The M.O. notes that in the two non-comparative studies, one half of cefepime patients with normal baseline values developed hypocalcemia and hyperphosphatemia. Also, nearly half (47%) of the patients developed elevations in ALT with 23% (9/39) deemed clinically relevant. The comparative studies did not show an imbalance between treatment groups for these electrolyte abnormalities. Worsening of laboratory parameters in the randomized trials of cefepime monotherapy was infrequent and occurred in less than one-quarter of the subjects (see Table 11.4 at the end of section 11). As with adverse clinical events, there were some differences between individual studies. The highest incidence was seen in study AI411-131. In most cases there was consistency across the four other trials. In comparing cefepime to ceftazidime, very few differences were detected between the two treatment groups. Alteration in renal function, characterized by an increase in either BUN/blood urea or serum creatinine, was similar in the two treatment groups. These changes were generally minimal and rarely required therapeutic intervention. Overall, the incidence of increases in BUN/blood urea was 15% in both groups, while increases in serum creatinine occurred somewhat more frequently in the ceftazidime group (8% versus 5%). Changes in liver enzymes were also evenly distributed between the two treatment groups and occurred in approximately 20% of the subjects overall. Increases in bilirubin were more frequent in the cefepime group (18% versus 8%). This difference was due in part to study AI411-189, where the respective incidences were 21% for cefepime and 7% for ceftazidime. Hypokalemia occurred in roughly one-third of the subjects, and were somewhat more frequent in the ceftazidime group. The changes were usually mild. In most instances, these changes in laboratory parameters could be attributed to the subjects’ underlying medical condition as well as to the ancillary therapies they received. This is particularly true for those subjects who received total parenteral nutrition, large volumes of IV fluids, and chemotherapeutic agents known for their liver toxicity. M.O. comment: In Study AI411-189, three times as many cefepime patients developed hyperbilirubinemia as in the ceftazidime arm, 21% versus 7%, respectively. Per the JCSR, “In most instances, these changes in laboratory parameters could be attributed to the subjects’ underlying medical condition as well as to the ancillary therapies they received. This is particularly true for those subjects who received total parenteral nutrition, large volumes of IV fluids, and chemotherapeutic agents known for their liver toxicity.”

In comparing cefepime to the combinations of gentamicin and either mezlocillin or piperacillin, differences between the treatment groups were ascribed to the use of an aminoglycoside and consisted primarily of electrolyte imbalances and alterations in renal function. Overall, the incidence of BUN abnormalities was 5% for cefepime and 15% for the combinations, while the incidence of increased creatinine was 6% and 26%, respectively. No differences were seen with respect to hyponatremia (cefepime 28% and combinations 24%), there was a larger proportion of subjects in the combination group who experienced hypokalemia compared to the cefepime group (43% versus 26%). Clinically relevant laboratory abnormalities were occasionally seen in the two non-comparative studies. The most frequent was an increase in ALT, which occurred in 9 (14%) of the 66 subjects who had normal baseline values. Clinically relevant hypocalcemia developed in one of ten subjects with normal baseline values. Other clinically relevant changes were usually seen in less than 5% of the subjects. Clinically relevant abnormalities were also infrequent in the comparative studies of cefepime monotherapy. In the comparison of cefepime to ceftazidime, no difference between the two treatment groups could be detected. The highest incidence of clinically relevant abnormalities was noted for hypophosphatemia, with an incidence of 8% in the cefepime group and 10% in the ceftazidime group. Clinically relevant elevations in bilirubin were noted in about 3-4% of the subjects in both groups. In the comparison of cefepime to combination therapy, the clinically relevant abnormalities follow the same pattern as in the comparison of cefepime to ceftazidime, with only one notable difference. In the combination group, 4 of 86 (5%) subjects developed clinically relevant increases in creatinine, compared to none in the cefepime group. As was

291

discussed earlier, this difference is likely related to the use of an aminoglycoside in the combination regimens. All other laboratory abnormalities were similar in the two treatment groups. Subjects with Abnormal Baseline Values In the non-comparative studies, worsening of laboratory parameters was infrequent in subjects with abnormal baseline values. With the exception of uremia and hyperbilirubinemia, they usually occurred in less than 25% of the subjects. There was a limited number of subjects with baseline laboratory abnormalities in the randomized trials of cefepime monotherapy. In the overall analysis of the worsening of laboratory values, there was little difference between the treatment groups. When present, the differences should be interpreted with caution due to the presence of pre-existing condition, as well as to the relatively small number of subjects in whom these abnormalities were described. The worsening to clinically relevant abnormalities in those subjects with abnormal baseline values is somewhat higher than that seen in subjects with normal baseline values. Overall, there was little difference between the two treatment groups for any of the parameters. There was no significant difference between treatment arms with respect time to recovery from neutropenia, as determined by Kaplan-Meier analysis (Figure 4). M.O. comment: According to the JCSR, there was no significant difference between treatment arms with respect to time to recovery from neutropenia. Discontinuation Due to Adverse Events In the non-comparative studies, discontinuation of study drug due to adverse events was relatively uncommon. In both AI411-143 (3 subjects) and AI411-158 (2 subjects), rash was the most common adverse event leading to discontinuation; in four instances, rash was felt to be probably drug related. Other events included drug fever, nausea, and urticaria. In the randomized trials of cefepime monotherapy, the development of a skin rash was by far the most common adverse event that led to study drug discontinuation. In the analysis of cefepime versus ceftazidime, this occurred in 5% of the subjects in the cefepime group and 3% in the ceftazidime group. Of note, rash was associated with fever in one ceftazidime subject and with fever, nausea, and vomiting in one cefepime subject. The comparison of cefepime to combination therapy is notable because of study AI411-137, in which a total of 24 subjects (10 cefepime, 14 gentamicin/mezlocillin) discontinued study therapy because of skin rash. All other adverse events leading to discontinuation occurred in 1% of subjects. Only one subject, in the ceftazidime arm of study AI411-189, discontinued study therapy because of a laboratory abnormality. In this instance, an elevated serum ALT was felt to be possibly related to ceftazidime and eventually resolved within four weeks of discontinuing drug. In the five comparative studies, 51 of 72 adverse events leading to discontinuation were felt to be probably drug-related; forty-three of these were rash. The other eight included fever, nausea/vomiting, and diarrhea. Other Studies Using Cefepime 2g q8h The majority of the clinical experience in febrile neutropenia subjects was accumulated in protocols in which cefepime was given at a dose of 2g q8h. In addition, 932 subjects enrolled in seven trials conducted in other indications on this dosage regimen. The vast majority of these subjects were accrued in a large multicenter trial conducted in the U.S. It was designed to compare cefepime given at a dose of 2g q8h to ceftazidime given at the same dose-schedule. A total of 421 subjects were accrued in the cefepime arm and 419 in the ceftazidime group. Additional experience came from six small non-comparative studies. In three studies, subjects were treated with several different dosing regimens, but data included in this report specifically focus on those who received cefepime at the dose of 2g q8h. The 60 subjects accrued in these 3 studies had a variety of infections, including lower respiratory tract infections (LRTI), urinary tract infections (UTI), and skin and soft tissue infections (SSTI). Seventeen subjects were accrued in a small non- comparative study of subjects with endocarditis and the remaining fifteen subjects were included in two emergency release protocols. Adverse events were reported infrequently in the 92 subjects included in the non-comparative studies of other indications. The most frequent event was rash, which was seen in 10 (11%) subjects. Other adverse events reported in 5% or more of subjects were headache, vomiting, nausea, and pruritus. There were very few adverse events of probable relationship to cefepime. These adverse events were usually mild to moderate in severity. The most frequent,

292

293

rash and pruritus, were reported in 4 subjects each. The other probably related adverse events seen in more than 2% of subjects were headache, dizziness, and taste alteration. All other events were described in single subjects. In the randomized trial of other indications (study AI411-160), no differences could be detected between cefepime and ceftazidime. The most frequent adverse events were diarrhea (13% vs 12%) and headache (11% in both arms). Other frequently reported adverse events were nausea, peripheral edema, and hyperventilation. For most events, no difference between cefepime and ceftazidime could be detected, although the incidence of peripheral edema was somewhat greater in the cefepime group (7% versus 4%). These events were generally not related to study drug. They may reflect differences in the overall condition of the subjects accrued in the trial. Most of these events were mild to moderate in severity; one cefepime subject developed a rash which was considered severe. Integrated Analysis of Adverse Events Analysis by Indication A total of 1,048 subjects were included in this integrated analysis of safety for cefepime administered at a dose of 2g q8h; 535 were included in the trials of febrile neutropenia and 513 in the trials of other indications. The most frequent adverse event was diarrhea, with an overall incidence of 14%, ranging from 17% in febrile neutropenia to 11% in other indications. Rash, gastrointestinal disturbances, and headache were the other frequently reported adverse events. Differences between indications were apparent for a number of other adverse events. For instance, rash was seen in 15% of the febrile neutropenic subjects compared to 5% of the remaining subjects. It can be speculated that this difference may be related to specific characteristics of the neutropenic subjects, such as reactions to radiation therapy, graft versus host disease (GVHD), or other toxic effects of high-dose chemotherapeutic agents. Less striking differences were seen for nausea and vomiting. For most of the other adverse events, there were limited differences between the indications. Analysis by Age Our database included data on 318 elderly subjects (age >65 years). Most of these elderly were included in the non- neutropenic trials. Overall, there was little difference between the two age groups, irrespective of diagnosis. Two adverse events should, however, be highlighted. Rash was less frequent in elderly subjects regardless of the indication. In contrast, confusion was four times more frequent in the elderly non-neutropenic subjects compared to their younger counterparts (9% versus 2%). M.O. comment: Of note, confusion was four times more frequent in the elderly non- neutropenic subjects compared to their younger counterparts. This may have been related to reduced renal clearance. Analysis by Gender The analysis by gender did not conclusively demonstrate a higher incidence of any adverse events in males or females. The only adverse event for which differences were identified was headache, with an overall incidence of 12% in females and 7% in males. This difference was seen in both neutropenic and non-neutropenic subjects. Analysis by Geographic Area Adverse events were more frequently reported in studies conducted in North America compared to those conducted in Europe. A direct comparison can only be made for febrile neutropenic subjects. In this subgroup, the incidence of diarrhea was 24% in the North American trials, compared to 9% in European studies. A similar difference was seen for almost all frequent adverse events. The difference described earlier between febrile neutropenic subjects and those treated for other indications was also seen if the analysis was restricted to North American trials. Subclassification of adverse events possibly related to the study drug by indication, age, gender or geographic area Limited data were available in subgroups when the adverse events that were possibly related to the study drug were considered. For this reason, discussion of this issue is of limited usefulness.

294

Table 11.2 All Adverse Events in Non-Comparative Febrile Neutropenia Trials

Number (%) of Subjects (N =114) Probably Unknown Not Related Total Related Any event 9 (8) 3 (3) 12 (11) 24 (21)

Rash 4 (4) 1 (1) 3 (3) 8 (7)

Diarrhea 3 (3) 2 (2) 2 (2) 7 (6)

Erythema 1 (1) --- 2 (2) 3 (3)

Nausea 1 (1) 1 (1) 1 (1) 3 (3)

Abdominal pain 1 (1) ------1 (1)

Vomiting ------1 (1) 1 (1)

Fever ------1 (1) 1 (1)

Mucositis ------1 (1) 1 (1)

Bradycardia ------1 (1) 1 (1)

Jaundice ------1 (1) 1 (1)

Lung infiltrate ------1 (1) 1 (1)

Angioedema ------1 (1) 1 (1)

Folliculitis ------1 (1) 1 (1)

Kidney ------1 (1) 1 (1) function abnormal

�

295

Table 11.3. Most Frequent Adverse Events in Phase III Febrile Neutropenia Trials Number (%) of Subjects

Cefepime Ceftazidime Cefepime Combination

131 189 204 Total 131 189 204 Total 118 137 Total 118 137 Total (N=45) (N=139) (N=143) (N=327) (N=45) (N=142) (N=133) (N=320) (N=59) (N=35) (N=94) (N=57) (N=36) (N=93) Any event 40 (89) 98 (71) 134 (94) 272 (83) 42 (93) 101 (71) 122 (92) 265 (83) 33 (56) 34 (97) 67 (71) 34 (60) 36 (100) 70 (75)

Rash 18 (40) 18 (13) 20 (14) 56 (17) 17 (38) 14 (10) 15 (11) 46 (14) 3 (5) 15 (43) 18 (19) 4 (7) 19 (53) 23 (25)

Diarrhea 18 (40) 16 (12) 23 (16) 57 (17) 14 (31) 11 (8) 30 (23) 55 (17) 7 (12) 19 (54) 26 (28) 11 (19) 27 (75) 38 (41)

Nausea 12 (27) 8 (6) 28 (20) 48 (15) 16 (36) 8 (6) 27 (20) 51 (16) 7 (12) 12 (34) 19 (20) 8 (14) 15 (42) 23 (25)

Abdominal pain 12 (27) 12 (9) 24 (17) 48 (15) 7 (16) 7 (5) 13 (10) 27 (8) 2 (3) 3 (9) 5 (5) 4 (7) 3 (8) 7 (8)

Vomiting 11 (24) 9 (6) 25 (17) 45 (14) 12 (27) 7 (5) 17 (13) 36 (11) 5 (8) 10 (29) 15 (16) 3 (5) 14 (39) 17 (18)

Headache 11 (24) 7 (5) 19 (13) 37 (11) 9 (20) 15 (11) 17 (13) 41 (13) 2 (3) 2 (6) 4 (4) 4 (7) 1 (3) 5 (5)

Fever 2 (4) 5 (4) 29 (20) 36 (11) 1 (2) 10 (7) 22 (17) 33 (10) 1 (3) 1 (1) 1 (2) --- 1 (1) --- Chills 13 (29) 1 (1) 23 (16) 37 (11) 10 (22) 1 (1) 16 (12) 27 (8) 5 (8) 3 (9) 8 (9) 2 (4) 8 (22) 10 (11)

Erythema 8 (18) 7 (5) 9 (7) 24 (7) 4 (9) 4 (3) 12 (9) 20 (6) --- 2 (6) 2 (2) 1 (2) 3 (8) 4 (4)

Edema 4 (9) 9 (6) 10 (7) 23 (7) 8 (18) 3 (2) 6 (5) 17 (5) --- 1 (3) 1 (1) --- 2 (6) 2 (2)

Hypotension 3 (7) --- 21 (15) 24 (7) 7 (16) 1 (1) 5 (/4) 13 (4) 3 (5) --- 3 (3) 1 (2) 1 (3) 2 (2)

Cough 2 (4) 11 (8) 11 (8) 24 (7) 4 (9) 3 (2) 9 (7) 16 (5) --- 1 (3) 1 (1) 1 (2) --- 1 (1)

Dyspnea 7 (16) 9 (6) 6 (4) 22 (7) 7 (16) 6 (4) 11 (8) 24 (8) 2 (3) 3 (9) 5 (5) --- 1 (3) 1 (1)

Pain 3 (7) 5 (4) 13 (12) 21 (6) 6 (13) 3 (2) 7 (5) 16 (5) 1 (2) 2 (6) 3 (3) 1 (1) --- 1 (1)

296

Table 11.3 (cont.) Number (%) of Subjects

Cefepime Ceftazidime Cefepime Combination

131 189 204 Total 131 189 204 Total 118 137 Total 118 137 Total (N=45) (N=139) (N=143) (N=327) (N=45) (N=142) (N=133) (N=320) (N=59) (N=35) (N=94) (N=57) (N=36) (N=93) Dizziness 9 (20) 1 (1) 11 (8) 21 (6) 5 (11) --- 10 (5) 15 (5) 2 (3) 1 (3) 3 (3) 2 (4) 3 (8) 5 (5)

Insomnia 4 (9) --- 15 (10) 19 (6) 3 (7) 1 (1) 20 (15) 24 (8) --- 3 (9) 3 (3) ------

Constipation 3 (7) 1 (1) 13 (9) 17 (5) 6 (13) 2 (1) 17 (13) 25 (8) 3 (5) 3 (9) 6 (6) 2 (4) --- 2 (2)

Petechiae 1 (2) 5 (4) 11 (8) 17 (5) 1 (2) 2 (1) 5 (4) 8 (3) ------

Asthenia 2 (4) 1 (1) 6 (4) 9 (3) 9 (20) 2 (1) 13 (10) 24 (8) --- 2 (6) 2 (2) 2 (4) 3 (8) 5 (5)

Peripheral edema 4 (9) 1 (1) 10 (7) 15 (5) 6 (13) 6 (4) 17 (13) 29 (9) 1 (2) 1 (3) 2 (2) ------

Confusion 4 (9) 2 (1) 9 (6) 15 (5) 8 (18) --- 5 (4) 13 (4) 2 (3) 4 (11) 6 (6) 1 (2) 4 (11) 5 (5)

Drowsiness --- 5 (4) 8 (6) 13 (4) 5 (11) 6 (4) 5 (4) 16 (5) --- 2 (6) 2 (2) 1 (2) 2 (6) 3 (3)

Lung disorder --- 1 (1) 9 (6) 10 (3) 2 (4) 3 (2) 16 (12) 21 (7) --- 2 (6) 2 (2) --- 2 (6) 2 (2)

Pain, mouth 3 (7) --- 12 (8) 15 (5) 4 (9) --- 4 (3) 8 (3) ------

Stomatitis 1 (2) 3 (2) 11 (8) 15 (5) ------4 (3) 4 (1) ------

* Occurring in 5% of cefepime or ceftazidime subjects in the pooled analysis.

297

Table 11.4. Laboratory Abnormalities in Cefepime Subjects with Normal Baseline Values Number of Subjects with Abnormal Values/Assessable Subjects (%) F/N Non- F/N vs. F/N vs. F/N Other Non- Other vs. Other Total Cefepime Comparative Ceftazidime Combinatio Total Comparative Ceftazidime Total 2g q8h n BUN 16/74 36/233 4/81 56/388 7/52 46/252 53/304 109/692 (22) (15) (5) (14) (13) (18) (17) (16) Creatinine 15/87 13/265 5/85 33/437 3/66 19/288 22/354 55/791 (17) (5) (6) (8) (5) (7) (6) (7) Alkaline phosphatase 18/69 42/190 13/61 73/320 11/40 41/214 52/254 125/574 (26) (22) (21) (23) (27) (19) (20) (22) AST/SGOT 24/80 34/189 8/59 66/328 9/40 67/224 76/264 142/592 (30) (18) (14) (20) (22) (30) (29) (24) ALT/SGPT 31/66 36/162 5/27 72/255 18/44 57/208 75/252 147/507 (47) (22) (19) (28) (41) (27) (30) (29) Bilirubin 12/54 34/191 8/63 54/308 2/51 14/273 16/324 70/632 (22) (18) (13) (18) (4) (5) (5) (11) Hyponatremia 20/63 44/183 14/50 78/296 7/46 59/267 66/313 144/609 (32) (24) (28) (26) (15) (22) (21) (24) Hypernatremia 4/63 7/183 2/50 13/296 1/46 12/267 13/313 26/609 (6) (4) (4) (4) (2) (4) (4) (4) Hypokalemia 21/67 68/233 21/82 110/382 10/67 52/302 62/369 172/751 (31) (29) (26) (29) (15) (17) (17) (23) Hyperkalemia 7/67 10/233 2/82 19/382 8/67 41/302 49/369 68/751 (10) (4) (2) (5) (12) (14) (13) (9) Hypocalcemia 5/10 54/130 0/39 59/179 3/38 66/159 69/197 128/376 (50) (42) (33) (8) (42) (35) (34) Hypercalcemia 0/10 1/130 0/39 1/179 1/38 6/159 7/197 8/376 (1) (1) (3) (4) (4) (2) Hypophosphatemia 2/10 44/167 12/42 58/219 8/46 61/201 69/247 127/466 (20) (26) (29) (26) (17) (30) (28) (27) Hyperphosphatemia 5/10 14/167 7/42 26/219 10/46 40/201 50/247 60/466 (50) (8) (17) (12) (22) (20) (20) (13)

298

RESISTANCE In the pooled evaluable monotherapy database (studies AI411-204 and 189), out of 99 isolates which had cefepime susceptibilities determined, 15 (15.2%) were resistant to cefepime. 109 isolates had susceptibility to ceftazidime determined; of these, 23 (21.1%) were resistant to ceftazidime. In all five monotherapy trials, the corresponding figures were 31/281 (11.0%) for cefepime and 47/291 (16.2%) for controls. These rates are consistent with previous literature on cephalosporin resistance. In general, treatment failures due to resistance were distributed evenly across treatment arms. In addition, secondary infections due to resistant organisms appeared at equal frequency in both cefepime and control arms. Thus, cefepime was equivalent to comparator regimens with regard to resistance, both with respect to in vitro results and clinical outcomes.

M.O. comment: The JCSR noted that treatment failures due to resistant pathogens were evenly distributed across treatment arms.

The following relates to the Labeling Recommendations made by the JCSR. LABELING RECOMMENDATIONS Given the results of the monotherapy studies, approval of the claim for effectiveness of cefepime in monotherapy of febrile neutropenic patients is recommended (see section 14). The applicant has proposed use of the word (b) (4) in the labeling for cefepime to describe the outcome of empiric therapy of febrile neutropenia. As discussed in the introduction to the reviews of the clinical studies, a number of outcome measures can be applied to assess the response to therapy, all of which may be regarded as (b) (4) Furthermore, use of the word (b) (4) carries a promotional connotation that should be avoided. For these reasons, a more neutral term such as (b) (4) ’ is recommended. With regard to the claim for effectiveness of cefepime as combination therapy, a regulatory action of non-approval is recommended, since the combination therapy studies failed to demonstrate therapeutic equivalence between cefepime combination therapy and the comparator regimen. However, complete absence of any reference to combination therapy in the label is problematic, since monotherapy is not appropriate for a number of patients at high risk for infection (e.g., patients with hematologic malignancies, patients with prolonged or severe neutropenia or patients with a history of bone marrow transplantation) or those with severe infection (e.g., patients with hypotension at presentation). The discussion at the March 5, 1997 AIDPAC meeting made clear in such patients, empiric monotherapy for febrile neutropenia would be unsafe. Given that this indication has never been granted before by the Division, it is important to have any label for this indication be as specific as possible, not only with respect to the intended usage but to special situations as well. For this reason, description of situations in which monotherapy may not be appropriate would be justified. It would be necessary, however, to indicate that there are insufficient data to demonstrate the efficacy of cefepime monotherapy in these situations.

M.O. comment: Per the JCSR, discussion during the March 5, 1997 AIDPAC meeting clarified that empiric antibacterial monotherapy would be unsafe for febrile neutropenia patients with high risk for infection (e.g., patients with hematologic malignancies, patients with prolonged or severe neutropenia or patients with a history of bone marrow transplantation) or those with severe infection (e.g., patients with hypotension at presentation).

Finally, inclusion of additional information in the form of a clinical study section giving information on the pooled monotherapy studies (AI411-189 and AI411-204) is warranted. Given that much of this information has not been published, description of the studies on which approval is based (demographics, response rates, and subgroup analyses) would provide useful information to prescribing physicians. Inclusion of such information, however, should not imply unsubstantiated claims. For example, description of the subgroup analyses should warn against drawing conclusions with respect to therapeutic equivalence between cefepime and ceftazidime, since these were post hoc analyses. The combination therapy studies (AI411-186 and AI411-198) should not be included in a clinical study section, since they do not represent adequate and well-controlled studies supporting the efficacy of cefepime for this indication.

299

A proposal for the Indication and Usage statement and a Clinical Study section is as follows. The data in Table 1 is derived by pooling demographic data on evaluable patients in studies AI411-204 and AI411-189; the data in Table 2 is derived from Table 10.2C.

M.O. comment: Per the JCSR, much of the clinical trial data related to the approval was unpublished.

INDICATION AND USAGE

Empiric therapy for febrile neutropenia. Cefepime is indicated for empiric monotherapy of febrile neutropenia. Antibiotic monotherapy may not be appropriate in patients at high risk for severe infection (including patients with a history of recent bone marrow transplantation, with hypotension at presentation, with an underlying hematologic malignancy, or severe or prolonged neutropenia). Insufficient data exist to demonstrate the efficacy of cefepime monotherapy in such patients (See CLINICAL STUDIES).

CLINICAL STUDIES

The safety and efficacy of empiric cefepime monotherapy of febrile neutropenia have been assessed in two multi-center randomized trials comparing cefepime monotherapy (at a dose of 2 g IV q8h) to ceftazidime monotherapy (at a dose of 2 g IV q 8 h). These studies included 317 evaluable episodes. Table 1 describes the characteristics of the evaluable patient population. Table 1. Demographics of evaluable patients Cefepime Ceftazidime Total 164 153 Median age (y) 56.0 (range, 55.0 (range, 18-82) 18-82) Male 86 (52%) 85 (56%) Female 78 (48%) 68 (44%) Leukemia 65 (40%) 52 (34%) Other hematologic malignancies 43 (26%) 36 (24%) Solid tumor 54 (33%) 56 (37%) Median ANC nadir (cells/µL) 20.0 (range, 0 20.0 (range, 0 500) 500) Median duration of neutropenia (d) 6.0 (range, 0 6.0 (range, 0 39) 32) Indwelling venous catheter 97 (59%) 86 (56%) Prophylactic Abx 62 (38%) 64 (42%) Bone marrow graft 9 (5%) 7 (5%) SBP <90 mm Hg at entry 7 (4%) 2 (1%) ANC, absolute neutrophil count; SBP, systolic blood pressure

M.O. comment: According to the previous table that combined evaluable patient data from Studies 189 and 204, there were 8% more patients with leukemia and other hematologic malignancies in the cefepime group compared to the ceftazidime group, 66% (108/164) and 58% (88/153), respectively. There were also 4% fewer solid tumor patients in the cefepime group compared to the ceftazidime group, 33% (54/164) and 37% (56/153), respectively. Such differences in baseline co-morbidity may have played a role, at least in part, in the increased mortality observed in the cefepime group.

300

Table 2 describes the clinical response rates observed. For all outcome measures, cefepime was therapeutically equivalent to ceftazidime.

Table 2. Pooled response rates for empiric therapy of febrile neutropenia % response Outcome measure Cefepime Ceftazidime (N=164) (N=153) Primary episode resolved with no treatment modification, no new 51% 55% febrile episodes or infection, and oral antibiotics allowed for completion of treatment Primary episode resolved with no treatment modification, no new 34% 39% febrile episodes or infection and no post-treatment oral antibiotics Survival of infection, any treatment modification allowed 93% 97% Primary episode resolved with no treatment modification and oral 62% 67% antibiotics allowed for completion of treatment Primary episode resolved with no treatment modification and no 46% 51% post-treatment oral antibiotics

Insufficient data exist to draw conclusions regarding the efficacy of cefepime monotherapy in patients at high risk for severe infection (including patients with a history of recent bone marrow transplantation, with hypotension at presentation, with an underlying hematologic malignancy, or severe or prolonged neutropenia). No data are available in patients with septic shock.

M.O. comment: In the pooled evaluable population for Studies 189 and 204, mortality was 4% higher in the cefepime group versus the ceftazidime group.

301

The following relates to the final recommendations made by the JCSR. RECOMMENDATIONS 1. The claim for effectiveness of cefepime as empiric monotherapy for febrile neutropenia at the proposed dosage of 2 g IV q8h is recommended for a regulatory action of approval, with appropriate labeling regarding lack of data regarding efficacy in patients at high risk for infection. 2. The claim of effectiveness of cefepime in combination with an aminoglycoside or glycopeptide for empiric therapy for febrile neutropenia is recommended for a regulatory action of non-approval. 3. A request for a phase IV commitment to a controlled trial comparing cefepime monotherapy to cefepime combination therapy is recommended.

M.O. comment: The M.O. notes that the Phase 4 commitment was not found in the May 15, 1997 approval letter sent to BMS.

7.4 Cefepime clinical trials included in the meta-analysis by Yu-te Wu, Ph.D.

The following Table 28 provides a list of the clinical trials included in the meta-analysis by Dr. Wu.

Table 28. Clinical Trials included in the Meta-analysis by Yu-te Wu, Ph.D.

302

303

304

M.O. comment: The M.O. defers the reader to the BMS submission dated 9/8/08 (revised “Master Table”) for a complete list of all the cefepime trials provided by BMS.

7.5 Line-by-Line Labeling Review

Please refer to Section 6.4 of this review.

305

7.6 Clinical Review of BMS’ submission entitled, “A Meta-Analysis of BMS Randomized Cefepime Trials”

Title of Submission: “A Meta-Analysis of BMS Randomized Cefepime Trials”

Submission and Review Dates Date of Submission: 3/13/09 Date Review Completed: 3/23/09, revised 4/6/09

Drug Identification Generic Name: Cefepime hydrochloride for Injection Trade Name: MAXIPIME

BACKGROUND In 2007, a meta-analysis was published by Yahav et al.1 that noted an increased risk of death among patients treated with cefepime versus other β-lactam agents. The Yahav meta-analysis was based on pooled study-level data from published papers, abstracts, and personal communications. The increased mortality was specifically noted among patients who received cefepime as empiric therapy for febrile neutropenia. In some of these studies, cefepime and the comparator therapy were combined with either an aminoglycoside or vancomycin.

Both the Agency and BMS moved forward with their own meta-analyses of mortality among patients treated with cefepime versus comparable agents. For further details regarding the Agency’s meta-analysis and overall analysis of mortality associated with cefepime, the M.O. refers the reader to (1) the Statistical evaluation of mortality risk associated with cefepime by Yu-te Wu, PhD, MPH (signed 1/30/09) and (2) other sections of this M.O. review.

BMS noted the following in the 3/13/09 submission, entitled, “A Meta-Analysis of BMS Randomized Cefepime Trials”.

“The purpose of the BMS meta-analysis was to explore the relationship between cefepime treatment and all-cause mortality using data from 40 randomized controlled trials supported by BMS (Appendix 1). All studies were completed before 2000. In addition to the meta-analysis, the current analysis utilizes patient-level data to explore the role of potential prognostic factors, i.e. Baseline Patient Characteristics, Therapy Administration, and Post-baseline Patient Factors. Analyses of prognostic factors are exploratory as the number of deaths in this database is too small for formal model validation.”

DESIGN & METHODOLOGY BMS noted the following on pages 8-11 of the submission.

Primary Objective “The primary objective is to explore the relationship between cefepime and all-cause mortality using a meta-analysis of data from randomized controlled BMS clinical trials. Reported deaths among treated subjects, regardless of cause or timing, are included.”

Secondary Objectives

306

“The secondary objectives are to: 1) Evaluate the effects of Baseline Patient Characteristics on mortality: • Evaluate the interaction between Therapy and Baseline Patient Characteristics on mortality. 2) Evaluate the effects of Post-baseline Patient Factors on mortality: • Evaluate the correlation between therapy and Post-baseline Patient Factors.”

Studies included in the BMS Meta-analysis “Analyses are performed using available data from 40 randomized clinical trials (total 9090 patients) sponsored by BMS. Specifications for these data were provided by the FDA (Appendix 2, Appendix 3). The adult portion and the pediatric portion of AI411131 are counted as separate trials. A list of the randomized trials, along with details on the number of subjects, comparators, indication, and mortality rates, is provided in Appendix 1.”

“For treatment assignment, subjects are either assigned to cefepime or comparator. All comparators from the different trials are combined. If a subject was re-randomized to several courses, as occurred in 4 Febrile Neutropenia studies (AI411131, AI411189, AI411198, and AI411204), the treatment assignment is determined by the last course.”

The following table, entitled, “Appendix 1: All-Cause Mortality by Study”, provides a list of the 40 studies included in the BMS meta-analysis. It was obtained from pages 41-42 of the submission.

307

308

M.O. comment: All of the studies included in the BMS analysis were also included in the FDA patient level and trial level analyses. In the FDA analyses, Study AI411131 was considered a single study and was not divided into two studies based on whether pediatric or adult patients were enrolled. There were slight differences in numbers of deaths per study because the BMS analysis counted all deaths post therapy, whereas, the FDA analysis only included deaths up to 30 days post therapy. The M.O. refers the reader to Appendix 7.4 of this

309 review to view the table of comparative clinical studies included in the FDA analysis performed by Dr. Yu-te Wu.

Primary Analysis: Cefepime versus All-cause Mortality “The principal meta-analysis is a Cochran Mantel Haenszel (CMH) odds ratio test, stratified by study, and tests the association between mortality and treatment (cefepime vs. comparator). The test includes all treated subjects from all trials. A subject is assumed to have died if death was reported in the study database; otherwise the subject is assumed to not have died. This includes subjects who died more than 30 days after the end of treatment, i.e. all-cause mortality. Treated subjects with missing death status at end of treatment are assumed to not have died. The CMH test performed is the standard test used by SAS v8, except that for studies without deaths in a treatment arm, a 0.5 correction was added to the number of deaths and non-deaths in each treatment arm. The standard CMH without the correction was done for sensitivity. The sensitivity analyses also include the risk ratio, risk differences, and an analysis modeling study as a random effect.”

“The primary and sensitivity analyses were repeated for the indications of Febrile Neutropenia and Multiple Infections, as these showed a potential for a difference between treatments. These subset analyses were only performed uncorrected since deaths occurred in each treatment arm in each study.”

Secondary Analysis #1: Baseline Characteristics & Therapy Administration “The complete list of Baseline Characteristics (including therapy administration) is shown in Table 3.3. The list contains the Characteristics and the range of values. Specifications for the derivation of these variables are in Appendix 2. Results are reported with an estimate, 95% Confidence Interval (CI), and p-values.”

“Each characteristic was screened individually using a logistic regression. Characteristics were analyzed adding a dummy parameter for each study unless the characteristic was strongly correlated with the different studies. Those characteristics associated (p < 0.05) with mortality and analyzed parameterized for study were selected to be analyzed for a treatment interaction. The selected Characteristics were modeled separately using a logistic regression model that included treatment and treatment interaction.”

The following table entitled, “Table 3.3: Marginal Baseline Factors for All-Cause Mortality”, was obtained from page 10 of BMS’ 3/13/09 submission.

310

Secondary Analysis #2: Post-baseline Factors “Post-baseline Factors were modeled dichotomously. A CMH test, based on odds ratio and stratified by study, was performed using each post baseline factor and death separately. A second CMH test was performed using each post-baseline factor and treatment separately. Analyses were uncorrected for zero deaths or zero factors as many factors occurred infrequently which would have resulted in a large number of corrections. The list of post-baseline patient factors analyzed is provided in Appendix 3.”

BMS also noted on page 11 that, “…the primary and sensitivity analyses were repeated for the indications of Febrile Neutropenia and Multiple Infections as these showed a potential for a difference between treatments.”

311

RESULTS Demographics The following table entitled, “Table 4.1: Overall Demographics”, was obtained from pages 12-13 of the submission and displays demographic information for the overall study population.

312

M.O. comment: In the “Overall” analysis, BMS included 5,095 cefepime patients and 3,995 comparator patients. In comparison, the FDA trial-level meta-analysis of 30-day mortality included 9,467 cefepime patients and 8,288 comparator patients, and the FDA patient-level meta-analysis included 5,058 cefepime patients and 3,976 comparator patients.

Regarding the BMS analysis, the two treatment groups were balanced in relation to the demographic variables studied. Of note, 65% of cefepime and comparator patients were enrolled at U.S. sites. Additionally, 4% more comparator patients were enrolled with renal- impairment as compared to cefepime patients.

The following table entitled, “Table 4.2: Febrile Neutropenia Demographics”, was obtained from pages 14-16 of the submission and displays demographic information for the febrile neutropenia study population.

313

314

315

M.O. comment: Regarding the BMS analysis, 57% of cefepime patients and 49% of comparator patients were enrolled at non-U.S. sites. BMS noted that the discrepancy was likely due to Study AI411186, in which, the randomization to cefepime + amikacin vs. ceftazidime + amikacin was 2:1. Additionally, more cefepime patients had hematologic malignancies and fewer solid tumors versus the comparator patients. Also, 8% more cefepime patients had a central line at baseline.

The following table entitled, “Table 4.3: Multiple Infections Demographics”, was obtained from pages 17-18 of the submission and displays demographic information for the “multiple infections” study population.

316

317

M.O. comment: In the “Multiple Infections” analysis, BMS included 1430 cefepime patients and 1436 comparator patients. In comparison, the FDA analysis included the “Multiple Infections” patients in an “Other” category that also included the following clinical conditions: bacterial meningitis, bacterial endocarditis, and bloodstream infections. The FDA trial level analysis of “Other” infections included 2,722 cefepime patients and 2,684 comparator patients, and the FDA patient level analysis included 2,327 cefepime patients and 2,216 comparator patients.

Regarding the BMS analysis, the two treatment groups were fairly well-balanced in relation to the demographic variables studied.

Dosing The following three tables entitled, “Table 4.4a: Dosing Overall”, “Table 4.4b: Dosing Febrile Neutropenia”, “Table 4.4c: Dosing Multiple Infections” were obtained from pages 19-20 of the submission and display dosing information for the three study populations.

318

319

M.O. comment: Based on BMS’ analyses, larger proportions of patients in the Febrile Neutropenia and Multiple Infections groups received 2 grams/dose as compared to the Overall population. Forty-seven percent of the Febrile Neutropenia (FN) patients received the labeled dosing regimen of three times/ day (TID). The M.O. notes that in Study AI411186, patients received cefepime 2 grams every 12 hours + amikacin, instead of the TID regimen. It is unclear why additional FN patients may have received less than TID cefepime dosing. One may speculate that at least a portion of those patients may have had impaired renal function that required modification of the cefepime dosing regimen.

All-Cause Mortality Analyses BMS provided all-cause mortality analyses (not limited to 30 days post therapy) for cefepime versus the comparator group based on crude mortality, Cochran Mantel-Haenszel (CMH) odds ratio, risk ratio, risk difference, and random effects odds ratio (corrected CMH). BMS provides their results for these analyses on pages 20-23. Tables 5.1.1a-5.1.4 were obtained from BMS’ submission.

Table 5.1.1.a provides the crude overall all-cause mortality for cefepime- and comparator-treated patients.

Table 5.1.1.b provides the CMH Odds Ratio for all-cause mortality stratified by study both with and without a correction of 0.5 deaths added to study treatment arms where no deaths occurred. Based on this methodology, the difference in mortality was not statistically significant.

320

Table 5.1.2 provides the results of BMS’ analysis of all-cause mortality based on risk ratio instead of odds ratio. Based on this methodology, the difference in mortality was not statistically significant.

Table 5.1.3 provides the results of BMS’ analysis of all-cause mortality based on risk difference. Based on this methodology, the difference in mortality was not statistically significant.

321

Table 5.1.4 provides the results of BMS’ analysis of all-cause mortality based on a random effects model and using a corrected CMH odds ratio. Based on this methodology, the difference in mortality was not statistically significant.

M.O. comment: Based on BMS’ analyses, there was no statistically significant difference in mortality between cefepime- and comparator-treated patients.

Mortality by Indication The following table entitled, “Table 5.1.5: Mortality by Indication”, was obtained from page 24 of BMS’ submission.

M.O. comment: The crude mortality rates for both treatment groups were higher in the BMS analyses, as compared to the FDA analyses because BMS did not limit mortality to within 30 days post therapy. The FDA analyses were based on mortality within 30 days of study therapy. This time interval was chosen in the hopes of capturing all deaths that could potentially be related to study therapy, while excluding deaths that were most likely unrelated to study therapy.

322

The following table, entitled, “5.1.5.1: Odds Ratio for Febrile Neutropenia Patients”, was obtained from page 24 of BMS’ submission. Results were based on odds ratio, risk ratio, and risk difference. BMS noted that corrections for no deaths in a study treatment arm were not necessary because all studies had at least one death in each treatment arm.

M.O. comment: Though there was a higher incidence of mortality among cefepime-treated versus comparator-treated febrile neutropenia patients, the association between cefepime exposure and mortality was not statistically significant.

The following table, entitled, “5.1.5.2: Odds Ratio for Multiple Infection Patients”, was obtained from page 25 of BMS’ submission. Results were based on odds ratio, risk ratio, and risk difference. BMS noted that corrections for no deaths in a study treatment arm were not necessary because all studies had at least one death in each treatment arm.

M.O. comment: Though there was a higher incidence of mortality among cefepime-treated versus comparator-treated patients in the “Multiple Infections” category, the association between cefepime exposure and mortality was not statistically significant.

323

Analyses of Interactions between Baseline Factors and Study Treatment in the “Overall” Population The following table, entitled, “Table 5.1.2b: Baseline Factor/Treatment Interaction”, was obtained from page 29 of the submission.

M.O. comment: Based on BMS’ analysis in the Overall population, pathogen susceptibility- by-treatment interaction had the smallest p-value. BMS noted that this analysis was performed adjusting for treatment indication (clinical condition treated) instead of study because most studies had so few cases.

The following table, entitled, “Table 5.2.1c: Table of Pathogen Susceptibility/Treatment Interaction”, was obtained from page 30 of the submission and demonstrates mortality rates by susceptibility and treatment.

324

M.O. comment: Based on BMS’ analysis in the Overall population, more cefepime patients with a resistant pathogen at baseline died as compared to comparator patients. The FDA identified a similar difference in 30-day mortality between cefepime and comparator regimens. However, upon further review of the case report forms and datasets of cefepime and comparator patients with resistant pathogens who died, the M.O. noted several discrepancies. These discrepancies included: (1) there were disproportionately more resistant pathogens isolated from cefepime-treated patients; (2) pathogens were incorrectly listed as “resistant” to study therapy (based on comparison with the information noted in the CRFs); (3) in many cases, it was unlikely that the resistant pathogen isolated at baseline caused the patient’s death (e.g., several of these patients had negative repeat cultures prior to death; (4) it was often unclear if the isolated organism was truly a pathogen vs. colonizer; and (5) many patients appeared to die due to co-morbid conditions, e.g., metastatic cancer. In 11/08, FDA discussed these issues with BMS, and in 12/08 BMS responded noting that upon further inspection of the datasets, it was discovered that the datasets contained “dirty/missing” data and that, in general, there were more missing susceptibility data for patients treated with comparator agents. Please refer to section 4.112 of this review for further details.

There is no indication from BMS that these discrepancies were reconciled prior to performing the “patient susceptibility-by-treatment interaction” analysis.

Analyses of Interactions between Baseline Factors and Study Treatment in the “Febrile Neutropenia” Population The following table, entitled, “Table 5.2.2a: Baseline Factor/Treatment Interaction-Febrile Neutropenia Patients”, was obtained from page 31 of the submission.

325

M.O. comment: Based on BMS’ analysis in the Febrile Neutropenia population, more cefepime patients with a history of diabetes at baseline died as compared to comparator patients.

The following table, entitled, “Table 5.2.2c: History of Diabetes/Treatment Interaction-Febrile Neutropenia Patients”, was obtained from page 32 of the submission.

326

M.O. comment: Based on BMS’ analysis of patients with Febrile Neutropenia and diabetes at baseline, 29.4% (10/34) cefepime patients and 6.8% (3/44) comparator patients died. This was a sub-subset analysis with very few patients. The M.O. suggests caution in drawing conclusions based on this relatively small sample of patients. Additionally, this finding was not identified in the Overall population analysis.

Analyses based on Post Baseline Factors On page 35 of the submission BMS noted that regarding the Overall Patient analysis, “None of the post baseline factors showed a significant correlation with treatment (cefepime vs. comparator).” BMS noted a similar finding for Febrile Neutropenia Patient analysis on page 36. BMS noted that for the Multiple Infections Patients, the only post baseline factor significantly associated with cefepime treatment was neurologic disorders (cefepime 7.07% versus comparator 4.89%, p=0.0116). However, this association was not observed in the analyses of either the Overall or Febrile Neutropenia Patients.

DISCUSSION Yahav et al.1 found a statistically significant increased risk of death among patients treated with cefepime versus other β-lactam agents. Neither the BMS nor the FDA meta-analyses confirmed these findings. Of note, the Yahav meta-analysis contained trial level mortality data from 41 publications and patient level data from no publications; BMS’ meta-analysis included trial level and patient level data from 40 studies (5,095 cefepime patients and 3,995 comparator patients); and the FDA meta-analyses included trial level data from 88 studies (9467 cefepime-treated patients and 8288 comparator-treated patients) and patient level data from 35 studies (5058 cefepime-treated patients and 3976 comparator-treated patients). Given that the FDA trial level mortality analysis included all of the Yahav and BMS studies, plus additional studies not included in the BMS analysis, the M.O. believes that the it is most likely that the FDA meta analyses provided the most reliable estimates of mortality associated with cefepime versus comparator agents.

For further details regarding the Agency’s meta-analysis and overall analysis of mortality associated with cefepime, the M.O. refers the reader to (1) the Statistical evaluation of mortality risk associated with cefepime by Yu-te Wu, PhD, MPH (signed 1/30/09) and (2) other sections of this review.

RECOMMENDATION The M.O. recommends that the FDA rely on the Agency’s meta-analysis as it appears to be the most all-inclusive regarding known mortality data for cefepime in comparative clinical studies, and is therefore most likely to provide the most reliable estimates of mortality associated with cefepime versus comparator agents.

RECOMMENDED REGULATORY ACTION Please refer to Section 6 of this review for details regarding recommended regulatory action related to this issue.

327

8 REFERENCES

1 Yahav D, Paul M, Fraser A, Sarid N, Leibovici L. Efficacy and safety of cefepime: a systematic review and meta analysis. Lancet Infect Dis. 2007 May;7(5):338-48.

2 Hughes WT, Armstrong D, Bodey GP, Bow EJ, Brown AE, Calandra T, Feld R, Pizzo PA, Rolston KV, Shenep JL, Young LS. 2002 guidelines for the use of antimicrobial agents in neutropenic patients with cancer. Clin Infect Dis. 2002 Mar 15;34(6):730-51. Epub 2002 Feb 13

3 Bhat SV, Peleg AY, Lodise TP Jr, Shutt KA, Capitano B, Potoski BA, Paterson DL. Failure of current cefepime breakpoints to predict clinical outcomes of bacteremia caused by gram-negative organisms. Antimicrob Agents Chemother. 2007 Dec;51(12):4390-5. Epub 2007 Oct 15

4 Pizzo PA, Hathorn JW, Hiemenz J, Browne M, Commers J, Cotton D, Gress J, Longo D, Marshall D, McKnight J, et al. A randomized trial comparing ceftazidime alone with combination antibiotic therapy in cancer patients with fever and neutropenia. N Engl J Med. 1986 Aug 28;315(9):552-8.

5 Sanders JW, Powe NR, Moore RD. Ceftazidime monotherapy for empiric treatment of febrile neutropenic patients: a meta-analysis. J Infect Dis. 1991 Nov;164(5):907-16.

6 Liang R, Yung R, Chiu E, Chau PY, Chan TK, Lam WK, Todd D. Ceftazidime versus imipenem-cilastatin as initial monotherapy for febrile neutropenic patients. Antimicrob Agents Chemother. 1990 Jul;34(7):1336-41.

7 Paul M, Yahav D, Fraser A, Leibovici L. Empirical antibiotic monotherapy for febrile neutropenia: systematic review and meta-analysis of randomized controlled trials. J Antimicrob Chemother. 2006 Feb;57(2):176-89. Epub 2005 Dec 12.

8 Gomez L, Estrada C, Gomez I, et al. Cefepime plus amikacin versus piperacillin-tazobactam plus amikacin in the treatment of patients with fever and granulocytopenia. Proceedings of the 41st Interscience Conference on Antimicrobial Agents and Chemotherapy; Dec 16-19, 2001; Chicago, IL, USA. Abstract L-771.

9 Chandrasekar PH, Arnow PM. Cefepime versus ceftazidime as empiric therapy for fever in neutropenic patients with cancer. Ann Pharmacother. 2000 Sep;34(9):989-95.

10 Chang SC, Fang CT, Hsueh PR, Liu CJ, Sheng WH, Hsieh SM, Hung CC, Chen YC. Efficacy and safety of cefepime treatment in Chinese patients with severe bacterial infections: in comparison with ceftazidime treatment. Int J Antimicrob Agents. 1998 Aug;10(3):245-8.

11 Chuang YY, Hung IJ, Yang CP, Jaing TH, Lin TY, Huang YC. Cefepime versus ceftazidime as empiric monotherapy for fever and neutropenia in children with cancer. Pediatr Infect Dis J. 2002 Mar;21(3):203-9.

12 Huang CK, Chen YS, Lee SS, Lin WR, Tsai HC, Lin HH, Wann SR, Chen JY, Yen MY, Liu YC. Safety and efficacy of cefepime versus ceftazidime in the treatment of severe infections. J Microbiol Immunol Infect. 2002 Sep;35(3):159-67.

13 Jiang JH, Wang Y. Comparison of the efficacy and safety between cefepime and ceftazidime in the treatment of the moderate to severe low-respiratory tract infection. Chin J Clin Pharmacol Therapeut 2003; 8: 92–94. [written in Chinese]

14 Kieft H, Hoepelman AI, Rozenberg-Arska M, Branger JM, Voskuil JH, Geers AB, Kluyver M, Hart HC, Poest- Clement E, van Beugen L, et al. Cefepime compared with ceftazidime as initial therapy for serious bacterial infections and sepsis syndrome. Antimicrob Agents Chemother. 1994 Mar;38(3):415-21.

15 Lin JC, Yeh KM, Peng MY, Chang FY. Efficacy of cefepime versus ceftazidime in the treatment of adult pneumonia. J Microbiol Immunol Infect. 2001 Jun;34(2):131-7.

328

16 Cherif H, et al. A prospective, randomized study comparing cefepime and imipenem-cilastatin in the empirical treatment of febrile neutropenia in patients treated for haematological malignancies. Scand J Infect Dis. 2004;36(8):593-600.

17 Sanz MA, López J, Lahuerta JJ, Rovira M, Batlle M, Pérez C, Vázquez L, Julià A, Palau J, Gutiérrez M, Capote FJ, Ramos F, Benlloch L, Larrea L, Jarque I; Spanish PETHEMA Group. Cefepime plus amikacin versus piperacillin-tazobactam plus amikacin for initial antibiotic therapy in haematology patients with febrile neutropenia: results of an open, randomized, multicentre trial. J Antimicrob Chemother. 2002 Jul;50(1):79-88.

18 http://www.srga.org/eucastwt/MICTAB/MICcephalosporins html [Website of EUCAST (European Committee on Antimicrobial Susceptibility Testing) under the auspices of the ESCMID (European Society for Clinical Microbiology and Infectious Diseases)]; last accessed on 1/20/09.

19 http://www.fda.gov/cder/drug/early comm/cefepime htm [Website of the U.S FDA, Center for Drug Evaluation and Research, “Early Communication About an Ongoing Safety Review Cefepime (marketed as Maxipime)”]; last accessed on 2/20/09.

20 http://www.fda.gov/cder/drug/early comm/cefepime update 200805.htm [Website of the U.S FDA, Center for Drug Evaluation and Research, “Follow-up to the November 14, 2007, Communication about the Ongoing Safety Review of Cefepime (marketed as Maxipime)”]; last accessed on 2/20/09.

21 Bhat SV, Peleg AY, Lodise TP Jr, Shutt KA, Capitano B, Potoski BA, Paterson DL. Failure of current cefepime breakpoints to predict clinical outcomes of bacteremia caused by gram-negative organisms. Antimicrob Agents Chemother. 2007 Dec;51(12):4390-5. Epub 2007 Oct 15

22 Cherif H, et al. A prospective, randomized study comparing cefepime and imipenem-cilastatin in the empirical treatment of febrile neutropenia in patients treated for haematological malignancies. Scand J Infect Dis. 2004;36(8):593-600.

329

------This is a representation of an electronic record that was signed electronically and this page is the manifestation of the electronic signature. ------/s/ ------Peter Kim 4/7/2009 06:21:39 AM MEDICAL OFFICER

Sumathi Nambiar 4/7/2009 12:35:51 PM MEDICAL OFFICER Addendum to Medical Officer’s Safety Review of Mortality Associated with Cefepime signed into DFS on 4/7/09 NDA 50-679

On page 186 of the original review, Table 17, entitled, “Trial Level Analysis of Patient Numbers and Studies by Treatment Indication” was cut-off. The following provides the complete table.

Table 17. Trial Level Analysis of Patient Numbers and Studies by Treatment Indication

Indications Number of Number of subjects Studies Cefepime (%) Comparator (%) Febrile Neutropenia 24 2791(29.48%) 2658(32.07%) Intra-abdominal Infection 7 628(6.63%) 470(5.67%) Pneumonia 26 2228(23.53%) 1821(21.97%) Urinary Tract Infection 7 763(8.06%) 490(5.91%) Skin Structure Infection 2 335(3.54%) 165(1.99%) Other 22 2722(28.75%) 2684(32.38%) Total 88 9467 8288

Sumathi Nambiar 4/17/2009 08:46:18 AM MEDICAL OFFICER Second Addendum to Medical Officer’s Safety Review of Mortality Associated with Cefepime signed into DFS on 4/7/09 NDA 50-679

On pages 27, 186, and 227 of the Medical Officer’s safety review of mortality associated with cefepime, the M.O. in describing highlights from Dr Wu's review noted the following:

“She performed a trial level analysis of 30-day mortality on the 41 studies already referenced by Yahav et al.1 plus an additional 47 studies. These 88 studies included 9467 cefepime-treated patients and 8288 comparator-treated patients.”

These sentences should be replaced with the following:

"She performed a trial level analysis of 30-day mortality on the 41 publications (38 trials) already referenced by Yahav et al.1 plus an additional 50 trials. These 88 trials included 9467 cefepime-treated patients and 8288 comparator-treated patients."

1 Yahav D, Paul M, Fraser A, Sarid N, Leibovici L. Efficacy and safety of cefepime: a systematic review and meta-analysis. Lancet Infect Dis. 2007 May;7(5):338-48. ------This is a representation of an electronic record that was signed electronically and this page is the manifestation of the electronic signature. ------/s/ ------Peter Kim 4/21/2009 02:29:45 PM MEDICAL OFFICER

Peter Kim 4/21/2009 02:30:23 PM MEDICAL OFFICER

Sumathi Nambiar 4/21/2009 02:34:48 PM MEDICAL OFFICER