Immunohistochemical Analysis of P-Glycoprotein Expression in Breast Cancer: Clinical Correlations*

Immunohistochemical Analysis of P-glycoprotein Expression in Breast Cancer: Clinical Correlations*

DAVID A. DECKER, M.D.,t LAURA W. MORRIS, B.S.,$ ALLAN J. LEVINE, M.D.,§ JANE E. PETTINGA, M.D . , 11 JENNIPHER L. GRUDZIEN, B.S.,f and DANIEL H. FARKAS, Ph.D.**

Departments of Medicine,t Anatomic Pathology,t and Clinical Pathology,§ William Beaumont Hospital, Royal Oak, MI 48073 and Department of Biology, Oakland University,l! Rochester, MI 48309 and Department of Surgery, William Beaumont Hospital-Troy,U Troy, MI 48098 and Bio-Oncology Laboratory,** Pontiac, MI 48057

ABSTRACT Paraffin-embedded, formalin-fixed tissue (PEFFT) specimens from a subset of breast cancer patients were analyzed by immunohistochemistry to determine whether or not the presence of P-glycoprotein identified che motherapy resistance. Two antibodies, C219 (monoclonal) and Abl (poly clonal), demonstrated appropriate immunostaining. Retrospectively and prospectively, P-glycoprotein expression was determined from PEFFT of 20 breast cancer biopsies (19 patients with locally advanced or metastatic disease). Immunohistochemical staining was graded for number of positive cells (NO to N4) and intensity (10 to 13). The immunostaining N and I of both antibodies were similar. There was no correlation between N, (P = 0.13) or I, (P = 0.67) and chemotherapy response or between N, (P = 0.63) or I, (P = 0.89) and survival. Five patients had residual cancer at repeat biopsy after systemic chemotherapy for locally advanced disease. These specimens had similar N and I as the primary cancer. This assay accurately identifies P-glycoprotein expression in PEFFT and revealed no correlation between expression and chemotherapy response or survival.

* Send reprint requests to David A. Decker, M.D., Cancer Care Associates, P.C., 3535 West 13 Mile Road, Suite 707, Royal Oak, MI 48073.

52 0091-7370/95/0100-0052 $01.20 © Institute for Clinical Science, Inc. IMMUNOHISTOCHEMICAL ANALYSIS OF P-GLYCOPROTEIN IN BREAST CANCER 53 Introduction tion. Clinical protocols to address these issues are being developed and reported. Intrinsic or acquired resistance to These investigations include intensive breast cancer chemotherapy remains a chemotherapy with either bone marrow major obstacle. A common factor in non autografting/stem cell harvesting or hema responsiveness is multidrug resistance topoietic growth factors and utilization of (MDR) . 1,2 In humans, two closely related chemosensitization agents that can inhibit MDR genes have been identified, MDR1 the outward transport function of P-glyco- and MDR2. Only MDR1 has been linked protein, e.g., verapamil, cyclosporin, quini- to the multidrug resistance phenotype, dine, or tamoxifen.8,9,10,11,12,13 encoding a highly homologous integral An immunohistochemical assay for transmembrane protein, Mdrl (P-glyco- MDR expression in paraffin-embedded, protein ) . 3,4 The P-glycoprotein (P for per formalin-fixed tissue (PEFFT) is meability) has a molecular weight of reported and its application to identify a 170,000 daltons and consists of a hydro- subset of breast cancer patients resistant phobic region with six transmembrane to chemotherapy is attempted. domains arranged in pairs, and a hydro philic cytoplasmic region containing Material and Methods potential nucleotide binding sites.5 This glycoprotein functions as an energy dependent outward transport pump that Antibodies to P-glycoprotein that could decreases intracellular drug accumula be used in PEFFT were validated. Eight (xm thick tissues sections were affixed to tion .3 ,4 Laboratory tumors expressing the MDR phenotype are resistant to a broad glass microscope slides. Tissue sections range of antineoplastic agents derived were deparaffinized with xylenes and from natural products, e.g., the vinca rehydrated through descending saline alkaloids (vincristine, vinblastine), (phosphate buffered saline, PBS). Tissue anthracyclines (doxorubicin, daunorubi- sections were incubated 30 minutes at cin), and podophyllotoxins (VP-16). Fur room temperature in PBS containing 5 thermore, resistance to other classes of percent non-immune serum to block non antineoplastic agents, the antimetabolite specific binding. Antibody dilution was trimetrexate and the synthetic dihydroxy- made using PBS containing 1 percent anthracenedione derivative mitox- normal serum. All incubations with per antrone, may also be mediated by this formed in a humidified chamber to pre vent evaporation. Monoclonal antibodies m echanism .6,7 Identification of MDR1 expression C219 (lOjxg/section) and C494 (5jxg/ holds the promise of being be a predictor section) were used in conjunction with for clinical drug resistance. Patients with Target Unmasking Fluid* as per the manu tumors expressing MDR1 should be poor facturers recommended protocol. Other candidates for therapy with vinca alka primary antibodies were polyclonal loids, anthracyclines, or podophyllotox M drl,t diluted 1:1000, and monoclonal ins. These individuals may be better anti-epithelial membrane antigen,t candidates for either no chemotherapy, diluted 1:5. All primary antibody incuba high-dose chemotherapy to increase intracellular concentration, therapy with antineoplastic agents whose resistance is * Signet Laboratories, Inc., Dedham, MA 02026. t Abl, Oncogene Science, Inc., Uniondale, NY not mediated by MDR, or protocols 11554. designed to inhibit P-glycoprotein func t Biomeda, Foster City, CA 94401. 54 DECKER, MORRIS, LEVINE, PETTINGA, GRUDZIEN, AND FARKAS tions were 2 hours at room temperature. A series of 19 patients treated at Biotinylated secondary antibodies,§ William Beaumont Hospital, Royal Oak diluted 1:500, were incubated 30 min and Troy, Michigan, was studied. utes at room temperature. Colorimetric Patients were identified either retrospec detection was made using Vectastain tively or prospectively from a single ABC-Alkaline phosphatase and Substrate medial oncology practice between Octo Kits (#SK5200).§ Endogenous alkaline ber 1988 and followed until date of death phosphatase was blocked by the addition or if still alive on April 1, 1993. Patients of Levamisole11 diluted 0.25mG/mL to the were selected if they had a biopsied substrate solution. breast cancer available in archival paraf Replicated sections assayed without fin measurable cancer, and received che the addition of primary antibody were motherapy within two months of the included to evaluate non-specific bind biopsy. Primary breast cancers (T2—T4b) ing from secondary antibodies and kits. were assayed in 11 patients (table I). Two Normal human kidney control sections of these (cases 8 and 14) also had meta were included in each assay to monitor static cancer at the time of presentation substrate incubation times and antibody (Ml). Five patients with locally advanced performance. Sections were counter cancer (T4b) had a diagnostic biopsy with stained with Nuclear Fast Red, dehy Mdrl assay followed by two to six cycles drated with alcohol, cleared with xylenes of chemotherapy and then a repeat post and mounted with Permount. A normal chemotherapy biopsy or mastectomy human tissue survey was conducted to with Mdrl assay. One patient (case 12) determine whether or not these antibod had a locally advanced breast cancer ies would stain PEFFT in a similar pat treated with chemotherapy followed by a tern to that described for similar frozen negative mastectomy specimen. She tissue specimens. C494 demonstrated experienced a chest wall recurrence poor staining on human control kidney (case 19) about one year later (Ml) and and was therefore not utilized further. again recovered chemotherapy for the The polyclonal rabbit antibody and C219 recurrent disease. Nine biopsies were both demonstrated an immunostaining from metastatic disease; either chest wall pattern on PEFFT similar to frozen speci or lymph node recurrence. mens. There was immunostaining on the Patients received only systemic che apical surface of both small and large motherapy except one who received con intestine, parietal cells of the stomach, current chemotherapy with mitox- Kulchitsky cells in the bowel, proximal antrone, fluorouracil, leucovorin (NFL), renal tubular cells, bile canalicular cells, and tamoxifen (case 8 ). All p atien ts hepatocytes, and adrenal cortex. Staining received at least two cycles of chemo was observed on the plasma membrane therapy. Response criteria were deter and/or cytoplasmic Golgi. Tissues mined clinically in a standard manner by expected to be negative were indeed physical exam and/or radiograms: com negative. The polyclonal antibody plete response (CR), partial response resulted in better resolution with more (PR), 50 percent regression from the maxi uniform staining at a much lower dilution mum diameter of all the initial tumors, than C219. stable disease (STAB), no change, and progression (PROG). Survival was calcu lated from the date of biopsy to death or if still alive on April 1, 1993. The following § Vector Laboratories, Burlingame, CA 94010. 11 Sigma Chemical Company, St. Louis, MO chemotherapy programs were used: NFL 63178. (mitoxantrone, fluorouracil, leucovorin), IMMUNOHISTOCHEMICAL ANALYSIS OF P-GLYCOPROTEIN IN BREAST CANCER 55

TABLE I

Patients, Stage, Chemotherapy, Expression of P-glycoprotein, Response, and Survival

Case Stage3 Chemotherapy1’ Expression of P-giycoproteinF Response11 Survival Mdr V Mdr 19 Mdr P (Ab 1) (C219) (Ab 1)

1 T2N0M0 CDF NO 10 CR 559+ 2 M1 NFL N4 13 N4 12 CR 320+ 3 T4bN0M0 CDF N1 11 N1 11 N 4 12 CR 485+ 4 T4bN0M0 CDF N2I1 NO 10 N1 11 CR 1626 5 T4bN0M0 CDF N4I1 N1 11 CR 1018+ 6 M1 NFL N4I1 PR 309 7 T4bN0M0 CDF N2I1 N1 12 PR 776 8 T3N0M1 NFL N0I0 PR 293+ 9 T3N0M0 CDF N1 11 N1 11 PR 316+ 10 M1 NFL N4I1 N4I1 PR 320+ 11 M1 CDF N2I1 N1 11 PR 725 12 T4bN0M0 CDF N1 11 PR 524 13 M1 CDF N4I1 PR 267+ 14 T4bN0M1 NFL N1 11 STAB 213 15 T4bN0M0 CDF N1 13 N4I1 N4 13 STAB 299 16 T4bN2M0 NFL N1 12 N1 11 STAB 423+ 17 M1 CMFVP N4I1 N4I1 STAB 482 18 M1 CMFVP N1 11 STAB 412 19 M1 CDF N 0I0 STAB 221 20 M1 D N on PROG 221

a Cancer-staging classification system with T referring to the primary tumor size, N to the clinical status of regional lymph node metastasis, and M to the remote metastasis. b Chemotherapy programs: CDF = cyclophosphamide, doxorubicin, and fluorouracil; NFL = mitoxantrone, fluorouracil, and leucovorin; CMFVP = cyclophosphamide, methotrexate, fluorouracil, vincristine, and prednisone; D = doxorubicin. c N = percent positive cells; I = intensity cell staining; NO = no positive tumor cells; N1 =coexistence of positive and negative tumor cells up to 25% positive; N2 = 26 to 50% positive; N3 = 51 to 75% positive; N4 = 76 to 100% positive; ¡0 = no staining tumors; 11 = weakly staining; 12 = moderate staining; and 13 = intense staining. d Clinical response definitions: CR = complete response; PR = partial response; STAB = stable disease (no change); and PROG = progression of disease. e Survival in days from biopsy; + = indicates alive at time of report. ' Mdr 1 prechemotherapy: polyclonal antibody (Ab 1) assay. 9Mdr1 prechemotherapy: monoclonal antibody (C219) assay. h Mdr 1 immediately after chemotherapy: polyclonal antibody (Ab 1) assay. CDF (cyclophosphamide, doxorubicin, All breast tissues were stained with the fluorouracil), CMFVP (cyclophospha- polyclonal antibody (Abl) and selected mide, methotrexate, fluorouracil, vincris- specimens were C219 (table I). All tis- tine, prednisone), and D (doxorubicin), sues were also stained with a monoclonal Standard doses and schedules were epithelial membrane antibody (EMA). employed; American Joint Committee Immunostaining of each section was Staging criteria were used. evaluated blindly by one of the authors 56 DECKER, MORRIS, LEVINE, PETTINGA, GRUDZIEN, AND FARKAS (AL) without knowledge of the clinical ber (N) of pretreatment positive tumor status. Two parameters of staining were cells or intensity (I) of staining predicted used. The first parameter was the num responsiveness to chemotherapy. Com ber of P-glycoprotein positive tumor paring NO to N4 versus clinical response cells: NO, no positive tumor cells; Nl, criteria, i.e., CR, PR, STAB, PROG, (P = coexistence of positive and negative 0.13) and 10 to 13 versus the same criteria, tumor cells up to 25 percent positive; N2, (P = 67) demonstrated no statistical dif 26 to 50 percent positive; N3, 51 to 75 ferences. Likewise, the number (N) of percent positive; N4, 76 to 100 percent pretreatment positive tumor cells or positive. The second parameter was intensity (I) of staining did not identify a staining intensity; 1 0 , indicated no stain subset with poor survival. Comparing NO ing tumor cells; II, weakly staining; 1 2 , to N4 versus survival, (P = 0.63) and 10 to moderate staining; 13, intense staining. 13 versus survival, (P = 0.89) demon Statistical analysis was preformed with strated no statistical difference. StatXact* and Systat 5.01.t Kaplan-Meier None of our patients with locally estimations and the log rank exact test advanced disease had a pathologic com were used to compare the number of plete response. P-glycoprotein expres positive cells (N) or intensity of stain sion was determined in five patients with ing (I) with survival. The Jonckheere- the polyclonal antibody before and again Terpstra test was used to compare the after receiving chem otherapy (table I). number of positive cells (N) or intensity An increase in either the number of of staining (I) with response (CR, PR, staining cells (N) or intensity of stain STAB, PROG). ing (I) following chemotherapy was not obvious. Results Discussion Nineteen patients with archival breast biopsies and documented clinical The identification of multidrug resis response data were studied with a rabbit tance in patients with breast cancer holds polyclonal antibody (Abl); nine of these the promise for exciting new therapeutic were studied with monoclonal antibody, developments. For this promise to be C219 (table I). Nine patients had locally realized sensitive assays are needed advanced breast (T2-4b, any N, MO), two which correlate with clinical chemo had locally advanced disease with therapy resistance. To date, none of the metastasis (T3NOM1 and T4bNOMl), published studies has adequately identi and nine and recurrent metastatic cancer fied a group of resistant patients who after failing local therapy (Ml). One could then be placed on chemotherapy patient (case 1 2 ) initially had locally protocols designed to circumvent advanced breast cancer and later (case MDR.14’15’1617 The ultimate role of MDR 19) recurrent biopsied local disease. in clinical oncology will be difficult to No correlation between immunostain- define. It will be difficult to devise an ing and chemotherapy response or sur acceptable assay to identify MDR and vival was observed. A group resistant to then incorporate this assay into well chemotherapy based on immunostaining designed clinical trials. Furthermore, could not be identified. Neither the num there are multiple determinants of drug responsiveness besides MDR which need to be considered. Patients without * Cytel Software Corporation, Cambridge, MA 02140. MDR expression may still be resistant. t Systat Inc., Evanston, IL 60201. These determinants include altered lev IMMUNOHISTOCHEMICAL ANALYSIS OF P-GLYCOPROTEIN IN BREAST CANCER 57 els of topoisomerase II, increased levels doxorubicin, mitoxantrone, or vincristine of glutathione S-transferase, and other is uncommon. Only one patient was energy dependent membrane transport treated with single agent therapy. Fur complexes similar to MDR but function thermore, other studies attempting to ing in the absence of MDR mRNA correlate the expression of Mdrl with expression . 18,19,20 chemotherapy responsiveness have also In the present series of 19 patients used doxorubicin based polychemother whose drug responsiveness was estab apy programs. 16,17 lished, responders could not be distin The total number of patients and the guished from non-responders by deter number with chemotherapy resistant dis mining the level of Mdrl expression with ease in this series was small. Certainly, an immunochemical assay. It had been the potential for a Type II statistical error anticipated that either a large number of exists. Similar sample sizes have been positive cells (N) or high intensity of investigated and reported by others. 16,17 staining (I) would identify chemotherapy In this series, only one patient had imme resistance. There are several explana diate progression, and six were stable. tions for the inability to identify a subset Some of the stable patients may have had of patients with chemotherapy resis chemotherapy resistant disease, since tance. Possibly, our immunohistochemi- they were only stable for two cycles of cal assay failed to identify the presence of chemotherapy with a subsequent short Mdrl. This seems unlikely since the survival. The patient with progression polyclonal antibody proved to be consis (case 2 0 ) and some of the stable patients tent and reproducible in the validation had low N and low I. These resistant study. It accurately identified Mdrl in cases may represent breast cancers PEFFT known to express Mdrl. The whose resistance is characterized by a human breast tissue staining pattern was mechanism other than MDR expression. similar to other reported series using The converse is observed in case 2, PEFFT. The immunostaining pattern of where the assay predicted resistance but the two antibodies, Abl and C219, were the patient was sensitive. Even though both similar. our study number is small, those cases Most of our patients received combina where expression did not correlate with tion polychemotherapy programs. Some chemotherapy responsiveness seem sig of these programs have drugs whose nificant. Even with larger patient num resistance is not mediated by MDR. The bers, the chance of this assay being chemotherapy program CDF was given highly predictive for chemotherapy 11 times and CMFVP twice. Resistance responsiveness seems small. Further to doxorubicin and vincristine in these more, a large sample size is not available combinations is mediated by MDR, but at our institution and evidently also not at such is not necessarily the case for the others. 16,17 Our sample size was limited other drugs. It is possible that patients by our strict inclusion criteria. in our series with large number of posi Few clinical studies attempting to uti tive cells (N) or intensity of staining lize MDR as a marker for chemotherapy (I) who responded to chemotherapy resistance or to circumvent MDR in responded to cyclophosphamide and/or human breast cancer patients have been 5-fluorouracil. An ideal study design rep o rted .14,15,16,17 Three of these are would determine MDR, then treat the immunohistochemical studies with dif patients with only a single natural prod ferent tissue sources (frozen or PEFFT), uct, e.g., doxorubicin. However, in clini varying chemotherapy programs, and cal practice, therapy with single agent diverse definitions of positive histologi 58 DECKER, MORRIS, LEVINE, PETTINGA, GRUDZIEN, AND FARKAS cal staining in terms of both the number designed clinical trials are needed which of cancer cells (N) and intensity (I).15,16,17 correlate with sensitive and specific labo One study, using the C494 monoclonal ratory determinations of MDR. antibody (MoAb) on frozen sections, did report identification of a small group of Acknowledgments chemotherapy resistant patients charac This work was supported by grants from the terized by 75 percent of the specimens William Beaumont Hospital Research Institute and staining strongly (I) . 17 Another study that the von Reis Breast Cancer Research Fund. The used the C219 MoAb on PEFFT defined helpful assistance of David Kessel, Ph.D. is acknowl edged in the preparation of this manuscript. positive immunostaining as the number of positive cells (N), and did not utilize intensity (I) as a criteria. There was no References obvious correlation between pretreat 1. Gerlach JH, Kartner N, Bell DR, Ling V. Mul ment expression and response. Identifi tidrug Resistance. Cancer Surv 1986;5:25-46. cation of a chemotherapy resistant subset 2. Pastan I, Gottesman M. Multiple-drug resis tance in human cancer. New Engl J Med 1987; was not possible . 16 The third study (an 316:1388-93. abstract) used the JSB 1 MoAb and 3. Bell DR, Gerlach JH, Kartner N, Buick RN, reported the presence of Mdrl to be sig Ling V. Detection of P-glycoprotein in ovarian nificantly predictive for a poor response cancer: A molecular marker associated with multidrug resistance. J Clin Oncol 1985;3: to anthracycline chemotherapy . 15 There 311-5. has been another abstract with a semi- 4. Kartner N, Riordan JR, Ling V. Cell Surface quantitative MDR1 mRNA assay suggest P-glycoprotein associated with multidrug resis tance in mammalian cell lines. Science 1983; ing that early levels of gene expression 221:1285-8. are a predictive marker of tumor response 5. Chen CJ, Chin JE, Ueda K, Clark DP, Pastan I, to neoadjuvant chemotherapy . 14 Gottesman MM, et al. Internal duplication and homology with bacterial transport proteins in There have been two reported chemo the mdrl (P-glycoprotein) gene from multi therapy trials that did not measure MDR drug-resistant human cells. Cell 1987;47: but instead attempt to overcome chemo 381-9. 6. Assaraf YG, Molina A, Schimke RT. Sequential therapy resistance by the addition of amplification of dihydrofolate reductase and verapamil or verapamil and quinine . 12,21 multidrug resistance genes in Chinese hamster In a randomized trial comparing single ovary cells selected for stepwise resistance to the lipid-soluble antifolate trimetrexate. J Biol agent epirubicin with or without vera Chem 1989;264:18324-6. pamil, no improvement in response was 7. Dalton WS, Durie BGM, Alberts DS, Gerlach observed by the addition of verapamil. 12 JH, Cress AE. Characterization of a new drug In the other trial the addition of vera resistant human myeloma cell line which expresses P-glycoprotein. Cancer Res 1986;46: pamil and quinine circumvented resis 5125-30. tance to an doxorubicin based program .21 8. Peters WP, Ross M, Vredenburg JJ, Meisenberg Unfortunately, none of these studies has B, Marks LB, Winer E. High-dose chemo therapy and autologous bone marrow support clearly defined a clinical role for the as consolidation after standard-dose adjuvant determination of MDR or offered any therapy for high-risk primary breast cancer. J conclusions on whether reversal of MDR Clin Oncol 1993; 11:1132-43. 9. Reed E, Vaughan W, Schmidt-Pokomy K, is possible. Edwards B, Kessinger A. High-dose chemo The current authors have validated an therapy with peripheral stem cell rescue in immunohistochemical assay for the patients with metastatic breast cancer. Breast Cancer Res Treat 1992;23:134. expression of Mdrl in PEFFT. With this 10. Trump DL, Smith DC, Ellis PG, Rogers MP, assay, a chemotherapy resistant subset of Schold SC, Winer EP, et al. High dose oral breast cancer patients could not be iden tamoxifen, a potential multidrug-resistance- reversal agent: phase I trial in combination tified. If the clinical promise of MDR with vinblastine. J Natl Cancer Inst 1992;84: identification is to be realized, then well 1811-6. IMMUNOHISTOCHEMICAL ANALYSIS OF P-GLYCOPROTEIN IN BREAST CANCER 59

11. Sonneveld P, Durie BGM, Lokhorst HM, Marie motherapy resistance in locally advanced JP, Solbu G, Suciu S, et al. Modulation of mul- breast cancer Hum Pathol 1990;21:787-91. tidrug-resistance multiple myeloma by cyclo 17. Verrelle P, Meissonnier F, Fonck Y, Feillel V, sporin. Lancet 1992;340:255-9. Dionet C, Kwiatkowski F, et al. Clinical rele 12. Mross K, Bohn C, Edler L, Jonat W, Q uesser W, vance of immunohistochemical detection of Heidemann E, et al. Randomized phase II multidrug resistance P-glycoprotein in breast study of single agent epirubicin + /— verapamil carcinoma. J Natl Cancer Inst 1991;83;lll-6. in patients with advanced metastatic breast 18. Nakagawa M, Schneider E, Dixon KH, Horton cancer. Ann Oncol 1993;4:45-50. J, Kelley K, Morrow C, et al. Reduced intracel 13. Wishart GC, Harnett A, Kerr DJ, Paul J, lular drug accumulation in the absence of Machem MA, Soukop M, et al. A randomized P-glycoprotein (mdrl) overexpression in mitox- placebo controlled trial of quinidine as a resis antrone-resistance human MCF-7 breast can tance modulator in patients with advanced cers. Cancer Res 1992;52:6175-81. breast cancer treated with epirubicin. Proc Am 19. Matsuo K, Kohno K, Takano H, Sato S, Kiue A, Soc Clin Oncol 1993;12:58. Kuwano M. Reduction of drug accumulation and DNA topoisomerase II activity in acquired 14. Chevillard S, Vielh Ph, Pouillart P. Tumor teniposide-resistant human cancer KB cell Response of Breast Cancer Patients Treated by lines. Cancer Res 1990;50:5819-24. neoadjuvant chemotherapy may be predicted 20. Batist G, Tulpule A, Sinha BK, Katki AG, Myers by measuring the early level of MDR1 gene CE, Cowan KH. Overexpression of a novel expression. Proc Am Soc Clin Oncol 1993; 12: anionic glutathione S-transferase in multidrug 59. resistant human breast cancer cells. J Biol 15. Pozza F, Bevilacqua S, Cazzavillian S, Meli S, Chem 1986;261:15544-9. Gasparini G. P-glycoprotein Expression pre 21. Dalton W, Taylor C, Mosley K, Dorr R, Salmon dicts chemoresistance in advanced breast can S. Combination chemotherapy with cyclophos cer. Proc Am Soc Clin Oncol 1993; 12:65. phamide, vincristine, adriamycin, dexametha- 16. Ro J, Sahin A, Ro JY, Fritsche H, Hortobagyi G, sone, plus quinine and verapamil in patients Blick M. Immunohistochemical analysis of with advanced breast cancer. Proc Am Soc Clin P-glycoprotein expression correlated with che Oncol 1993; 12:63.