Correspondence: I. Potasman, MD Leukemoid Reaction: Spectrum and Prognosis of 173 Adult Patients Infectious Diseases, ABSTRACT No. Bnai Zion Med. Ctr. 47 Golomb St., Haifa 31048. 39147 Israel Potasman, MD and Moti Grupper, MD Tel. 972-48359055 Fax. 972-48359755 Bnai Zion Med. Ctr., Haifa, Israel Email: [email protected]

LR- CLINICAL and LABORATORY FEATURES CHRONIC DISEASES & CONDITIONS CAUSING LR METHODS WikipediA: The term leukemoid reaction describes an elevated white blood

cell count, or , that is a physiological response to stress or infection (as Table 1: Leukemoid Reaction: Demographics, Major Clinical, and Laboratory features, and Table 2: Chronic & possible Instigating diseases Causing Leukemoid Reaction (n=173)  The BZMC is a 411 bed university hospital located in Haifa, Israel. Mortality (n=173) Disease/Drug Occurrences (%) opposed to a primary blood malignancy, such as ).  Contains all basic departments except for neurosurgery, cardiovascular surgery and solid-tumor . The respi- Parameter N (%) Alive Dead n Age (years, mean ±S.D.) 69.4 ±19.6 63.3±21.2 79.4±11.2 Background ratory intensive care unit inhabits six patients; additional respirators are in use in the 3 departments of medicine. History of Cardiovascular 86 (49.7%) Minimal-Maximal 21-97 21-97 28-97 ABSTRACT  During an average year there are 29,508 hospitalizations of adults (>18 years) with ~150,000 hospitalization-days, Median 75 69 81 disease Diabetes Mellitus 46 (26.6%) and an average occupancy of 92%. Patients were eligible throughout hospital stay, and wherever they were admit- Chronic Lung disease 29 (16.8%) ted. Objectives: The prognosis of patients with Leukemoid Reaction (LR) depends mainly on their underlying illness. Our aim was Sex (m/f) 85/88 Instigating conditions* Acute renal failure 24 (13.9%) Admitted from  The study spanned from March 2011 through February 2012: At first, the hospital's laboratory's com- to investigate the etiologies and prognosis of a mixed group of patients with LR. Tumor (inactive) 22 (12.7%) 120 (69.4) 84 (48.6) 36 (20.8) -Home (%) Acute 13 (7.5%) 9 9 puter was searched for all blood counts with a cutoff of ≥30.0 X10 leukocytes/µL, and with a count of Design/subjects: We identified 173 patients who had ≥30.0 X10 leukocytes/µL without hematologic malignancies. Parame- -Nursing home (%) 53 (30.6) 23 (13.3) 30 (17.3) Vaginal delivery 6 (3.5%) Splenectomy 5 (2.9%) >50%. Children ≤18 years and were excluded. Diagnostic Groups ters suspected of causing LR and factors contributing to death were analyzed. Cesarean section 3 (1.7%)  -Infections n, (%): 83 (48) 44 (25.4) 39 (22.5) ketoacidosis 3 (1.7%) The search yielded 218 records. 45 patients were excluded because of hematologic malignancies. Thus, this study Results: Patients with LR constituted 0.59% of all admitted adults. The median age was 75 years, but twenty were under 40. # Eclampsia, hemolysis, 0 sepsis 16 (9.2) 5 (2.9) 11 (6.4) depicts the details of 173 (79.3%) individual patients. 9 poisoning There was no difference in LR prevalence by gender (F/M=88/85). Average WBC count was 37.7 X10 /µL. Fourteen patients pneumonia 15 (8.7) 8 (4.6) 7 (4) Drugs:  Admission diagnoses contained 9 major categories: sepsis (defined by previously established criteria), pneumonia, (8.0%) had a WBC count of >50.0 X109/µL. The median duration of LR was 1 day, but 39 patients had prolonged LR (>1day). urinary tract infection 14 (8.1) 11 (6.4) 3 (1.7) Steroids [n>20mgs] 18 [11] 10.4% UTI, other infections (diarrhea, cellulitis, etc.), obstetrics/ gynecology (mostly labor), tissue ischemia/stress, other: diarrhea, cellulitis, gangrene 38 20 (11.6) 18 (10.4)

Infection was the most common cause of LR (n= 83, 47.9%; 95% CI=40.7-55.4), followed by ischemia/stress (27.7%), inflam- inflammation (e.g. pancreatitis, splenectomy) solid tumors and "other". -Tissue ischemia/physiological stress* 48 (27.7) 31 (17.9) 17 (9.8) Adrenalin 8 (4.6%) Non-steroidal anti- 5 (2.9%) mation (6.9%) and obstetric diagnoses (6.9%). Higher WBC counts were significantly associated with positive blood cultures (p -Inflammation** 12 (6.9) 8 (4.6) 4 (2.3)  Statistical analysis has been executed using SPSS software (2009). The study was approved by the IRB of the BZMC. inflammatory - 12 (6.9) 12 (6.9) 0 = 0.017), or a positive C. difficile toxin (p = 0.001). Antibiotics were prescribed to 140 patients (80.9%). Obstetric/gynecology Carbamazepine 2 (1.2%) -Malignant tumor*** (active) 8 (4.6) 7 (4.0) 1 (0.6) Minocycline, lithium 0 RESULTS

Sixty six patients (38.1%) died during hospitalization. Those with prolonged LR had an in-hospital mortality rate of 61.5%. Fac- 10 (5.8) 5 (2.9) 5 (2.9) -Miscellaneous *Additional causes for LR could not be excluded We identified 173 patients with LR, constituting 0.59% of all admitted adult patients. Positive blood cultures§ 22 (12.7) 8 (4.6) 14 (8.1) tors found to be highly correlated with death were: age (OR = 1.051, p < 0.001), any infectious diagnosis (OR = 2.574, p=0.014)  This cohort was relatively old with a median age of 75; however, 20 patients were under 40 years. Had diarrhea§ 21 (12.1) 10 (5.8) 11 (6.4) 9 9 Positive Clostridium difficile toxin assay§§ 12 (6.9) 3 (1.7) 9 (5.2)  The average WBC count was 37.7 X10 /µL (Table 1), with a maximal count of 88.0 X10 /µL. Fourteen patients (8.0%) and sepsis (OR = 3.752, p = 0.001). Table No. 3: Abbreviated Logistic Regression of Predictors of Death Invasive procedures§ 54 (31.2) 9 had an extreme WBC count of >50.0 X10 /µL (of these, five had infections, while others suffered from inflammatory, Received antibiotics 140 (80.9) Conclusions: Leukemoid reaction carries a grave prognosis, especially among the elderly, and those with sepsis. LR was found ischemic and extreme physiological stress conditions). Died during hospitalization 66 (38.1) Parameter OR 95% CI p value Mean maximal WBC (X109/L ±S.D.) 37.7 ±8.8 36.6 ±7.1 39.5 ±11  In most patients (n=134) LR was a single-day event. However, in 39 (22.5%) the LR has continued for up to 17 days to have multiple etiologies including infections, stress, inflammation and obstetric diagnoses. Adrenalin Mean duration of LR [days (±S.D.)] 1.7 ±2.1 9.37 1.04-84.18 0.046 (median=2 days; mean=3.9 days; interquartile range=2). Age 1.05 1.02-1.08 0.000  While only 83 patients had an infection, 140 received antibiotics; Median duration of WBC>30.0 (X109/L) 1 INTRODUCTION Any Infectious Mean days post admission of Max. WBC 4.8 ±12 2.3 ±3.2 8.9 ±18.5 2.57 1.21-5.47 0.014  Sixty six patients of this cohort (38.1%) died during hospitalization, in contrast with a hospital-wide mortality rate of Diagnosis 5.71% (557/9750) among those aged >70 years (p<0.0001). In-hospital mortality rate was influenced by several fac- The term Leukemoid Reaction (LR) was coined by Krumbhaar in 1926. Since then, it has been clear that this term should be re- Sepsis 3.75 1.776-7.98 0.001 tors, notably older age (see Table 3). served for cases with non-hematologic malignancies. Yet the exact definition of LR is unsettled. While some authorities have Constant # Five patients with sepsis had bacteremia 0.005 1.12 0.000 9 9 9 9 CONCLUSIONS used a cutoff of 25.0 X10 leukocytes/µL, others have applied a cutoff of 30.0 X10 /µL, 40.0 X10 /µL, or even 50.0 X10 /µL. Per  Includes: Cardiopulmonary distress (n=22), abdominal pain/bowel References: definition, the elevated ’ count of LR is accompanied by a left shift, signs of neutrophils' activation, and absence of obstruction (n=12), gastrointestinal bleeding (n=6), major operations. Leukemoid reaction carries a grave prognosis, especially among the elderly (>60), and 1. Krumbhaar EB– Am J Med 1926. and dysplastic changes. The range of conditions causing LR is dependent primarily on the spectrum of patients stud- ** Includes: pancreatitis (n=5), splenectomy (n=5), cholelithiasis, diverticulitis 2. Lawrence YR– QJM 2009. those with any infection, but especially sepsis. LR was found to have multiple etiologies ied. Reding (1998) has studied mostly male patients, while Granger (2009) has focused on nonhematologic patients. without perforation. 3. Holland SM– Harrison’s 2008. *** Other causes for LR were excluded 4. Granger JM– Cancer 2009. including infections, stress, inflammation and obstetric diagnoses. With the wisdom of § All these parameters occurred just before, or concurrently with the LR. 5. Halkes CJ– Neth J Med 2007. As a tool for clinicians we report the etiologies and prognostic factors of LR of 173 pts without hematologic malignancies. 6. Sakka V– Eur J Intern Med 2006. hindsight, antibiotics had been overprescribed in this cohort. §§ Seven of 12 patients in this group had fever, but none had acute renal failure. 7. Reding MT– Am J Med 1998.