Etiology and outcome of extreme leukocytosis in 758 nonhematologic cancer patients

A retrospective, single-institution study


  • John M. Granger MD, MPH,

    1. Department of Infectious Diseases, Infection Control and Employee Health, The University of Texas M. D. Anderson Cancer Center, Houston, Texas
    Search for more papers by this author
  • Dimitrios P. Kontoyiannis MD, ScD

    Corresponding author
    1. Department of Infectious Diseases, Infection Control and Employee Health, The University of Texas M. D. Anderson Cancer Center, Houston, Texas
    • Department of Infectious Diseases, Infection Control and Employee Health, Unit 402, The University of Texas M. D. Anderson Cancer Center, 1515 Holcombe Boulevard, Houston, TX 77030===

    Search for more papers by this author
    • Fax: (713) 745-6839



To the authors' knowledge, the literature regarding extreme leukocytosis in solid tumor patients is sparse, consisting of a few case reports and small case series.


A total of 3770 consecutive solid tumor patients with a white blood cell count>40,000/μL were retrospectively identified over a 3-year period (2005-2008). Those patients without a secondary cause of their leukocytosis were defined as having a paraneoplastic leukemoid reaction.


A total of 758 (20%) patients with solid tumors and extreme leukocytosis were identified. The etiology of the leukocytosis was hematopoietic growth factors in 522 (69%) patients, infection in 112 (15%) patients, high-dose corticosteroids in 38 (5%) patients, newly diagnosed leukemia in 9 (1%) patients, and paraneoplastic leukemoid reaction in 77 (10%) patients. The patients diagnosed with a paraneoplastic leukemoid reaction typically had neutrophil predominance (96%) and radiographic evidence of metastatic disease (78%), were clinically stable, and had a poor prognosis; 78% either died or were discharged to hospice within 12 weeks of their initial extreme leukocyte count. All of the 8 (10%) patients who survived>1 year received effective antineoplastic therapy.


Infection was an uncommon cause of extreme leukocytosis in patients with solid tumors. Patients with paraneoplastic leukemoid reactions typically were clinically stable despite having large tumor burdens. However, clinical outcomes were poor unless effective antineoplastic treatment was received. Cancer 2009. © 2009 American Cancer Society.

Extreme leukocytosis in solid tumor patients, variably defined as a white blood cell (WBC) count ≥40,000/μL or 50,000/μL, has been reported in nearly all solid tumor types and is historically associated with poor outcomes.1-3 Nonclonally derived mature neutrophils usually drive the leukocytosis.1 To our knowledge, its frequency in patients with nonhematologic cancer remains unclear, having been reported to range from 1% to 4% in small case series.1, 4 Extreme leukocytosis represents a diagnostic dilemma because of the need to rule out such secondary causes as infections, newly developed hematologic malignancies, use of corticosteroids or hematopoietic growth factors, severe hemorrhage, and metastases to bone with necrosis.5 Paraneoplastic leukemoid reaction (PLR) remains a diagnosis of exclusion.

To the best of our knowledge, PLRs have been insufficiently characterized.2, 4, 6 Affected patients typically appear clinically stable despite having metastatic cancer or a large tumor burdens.3, 7, 8 Elevated serum cytokine levels have been reported in these patients, chiefly granulocyte–colony-stimulating factor (G-CSF),9-11 although elevated serum levels of granulocyte-macrophage–colony-stimulating factor (GM-CSF),10, 11 interleukin-1α,6 and interleukin-69, 12 have also been measured. Elevated serum levels of multiple cytokines have also been documented.6 These cytokines may also promote tumor growth in a paracrine manner.13, 14 In addition, the transplantation of tumor cells into mouse models induced neutrophilia driven by leukogenic cytokines produced by tumor cells.15, 16 However, there are reports in which no serum cytokines were found to be increased in patients with a PLR.17 PLRs typically occur late, usually a few weeks or months before death.8, 18 There are, however, a few case reports of patients achieving resolution of their leukocytosis with the initiation of effective antineoplastic therapy.9, 12

In this study, we reviewed our recent experience with extreme leukocytosis in a large number of patients with solid tumors.


Using electronic medical records, we identified all patients treated in our tertiary cancer center who had a leukocyte count ≥40,000/μL (January 1, 2005 through February 20, 2008). If a patient had multiple high leukocyte counts, only their first high count was considered. After excluding all patients with a known pre-existing hematologic malignancy, we retrospectively collected vital signs and clinical data and made a determination of the etiology of the extreme leukocytosis. A patient's leukocytosis was determined to be due to the use of CSFs if these agents had been used within the prior 10 days. Leukocytosis was attributed to infection when there was a positive culture from a normally sterile body site or a chest radiograph suggestive of pneumonia and clinical improvement along with reduction of leukocytosis as a response to appropriate antimicrobial therapy. A high leukocyte count was attributed to the use of high-dose corticosteroids and/or vasopressors if the patient had not recently received CSFs and had no radiographic or microbiologic evidence of infection and if the leukocytosis improved with tapering of the above-mentioned agents. New diagnoses of hematologic malignancies were based on bone marrow biopsy. All remaining patients were defined as having PLRs. Tumor burden was indicated by diagnostic imaging, but we did not independently stage the different tumor types. Patient outcomes were dated starting from the first extreme WBC count.


Of the 3770 consecutive patients with a leukocyte count >40,000/μL, 758 had solid tumors (20%). In these 758 patients, leukocytosis was attributed to hematopoietic growth factors in 522 (69%) patients, infection in 112 (15%) patients, high-dose corticosteroids and/or vasopressors in 38 (5%) patients, newly diagnosed leukemia in 9 (1%) patients, and PLR in 77 (10%) patients.

Of the 112 patients with documented infections, 55 (49%) were diagnosed with pneumonia (Table 1). Bloodstream infections were documented in 30 (27%) patients. Urinary tract infections, peritonitis, and wound infections were noted in 18 (16%) patients, 15 (13%) patients, and 15 (13%) patients, respectively. Three (3%) patients presented in apparent septic shock, although their blood cultures were negative. One (1%) patient each had colitis and meningitis. Twenty-seven (24%) patients had multiple sources of infection. The infectious agents were diverse, with no 1 specific species or class of pathogens predominating.

Table 1. Sources of Infection in Solid Tumor Patients With Extreme Leukocytosis (n=112)
Etiology of InfectionNo. of Patients (%)
  • BAL indicates bronchial–alveolar lavage; UTI, urinary tract infection.

  • *

    Cytomegalovirus (CMV) in 3 cases, Streptococcus pneumoniae in 2 cases, Aspergillus fumigatus in 1 case, Fusarium species in 1 case, methicillin-resistant Staphylococcus aureus (MRSA) in 1 case, and Klebsiella pneumoniae in 1 case.

  • Polymicrobial in 7 cases, Klebsiella pneumonia in 9 cases, Pseudomonas species in 4 cases, coagulase-negative staphylococci in 4 cases, MRSA in 3 cases, Candida albicans in 3 cases, Enterococcus spp. in 3 cases, methicillin-sensitive Staphylococcus aureus (MSSA) in 2 cases, C. glabrata in 2 cases, Escherichia coli in 2 cases, Pantoea agglomerans in 1 case, C. parapsilosis in 1 case, Acinetobacter spp. in 1 case, Citrobacter koseri in 1 case, Cryptococcus neoformans in 1 case, Bacteroides fragilis in 1 case, and α-hemolytic Streptococcus in 1 case.

  • Polymicrobial in 4 cases; Escherichia coli in 5 cases; Klebsiella pneumoniae in 3 cases; yeast, not further identified in 3 cases; MRSA in 2 cases; Candida glabrata in 1 case; C. parapsilosis in 1 case; C. tropicalis in 1 case; Pseudomonas spp. in 1 case; Proteus mirabilis in 1 case; and clinical diagnosis, no culture done in 1 case.

  • §

    Clinical/radiographic diagnosis in 9 cases, polymicrobial in 3 cases, Candida albicans in 3 cases, C. glabrata in 2 cases, coagulase-negative staphylococci in 2 cases, Peptostreptococcus spp. in 1 case, α-hemolytic Streptococcus in 1 case, Stenotrophomonas maltophilia in 1 case, Klebsiella oxytoca in 1 case, Enterococcus spp. in 1 case, C. lusitaniae in 1 case, and C. tropicalis in 1 case.

  • Clinical diagnosis in 6 cases; polymicrobial in 8 cases; Enteric gram-negative rods, not otherwise identified in 6 cases; MRSA in 4 cases; Enterococcus spp. in 3 cases; Candida albicans in 3 cases; Pseudomonas aeruginosa in 2 cases; α-hemolytic Streptococcus in 2 cases; β-hemolytic Streptococcus in 1 case; Candida non-albicans spp. in 1 case; coagulase-negative staphylococci in 1 case; Proteus spp. in 1 case; Citrobacter freundii in 1 case; Peptostreptococcus spp. in 1 case; and Bacteroides fragilis in 1 case.

  • Clinical/radiographic diagnosis.

  • #


Pneumonia55 (49)
 Diagnosed by radiology46
 Diagnosed by positive BAL culture*9
Bloodstream infection30 (27)
UTI18 (16)
Peritonitis§15 (13)
Wound abscess13 (12)
Sepsis, without diagnostic culture or radiograph3 (3)
Colitis1 (1)
Meningitis#1 (1)
Multiple sources of infection27 (24)

The 77 solid tumor patients defined as having a PLR were examined further. There was a diversity of underlying solid tumors (Table 2). Nearly all patients had large tumor burdens (either a large bulky primary tumor or widely metastatic tumors); 60 (78%) patients had radiographic evidence of multiple metastases. Forty-one patients (53%) had evidence of lung involvement. The mean WBC count was 53,000/μL (range, 40,600/μL −115,100/μL) and the vast majority of patients (74 patients; 96%) had neutrophil predominance. Of the 49 patients for whom data were available, 34 (69%) had a WBC >20,000/μL at least 1 month before their first recorded extreme leukocyte count (WBC ≥40,000/μL). Of the 76 patients with a PLR for whom data were available, the majority were clinically stable at the time of presentation: 99% (75 patients) were afebrile, 93% (71 patients) were normotensive, and 88% (67 patients) had a normal respiratory rate, although a slight majority (53%, 40 patients) had tachycardia.

Table 2. Demographics, Clinical Characteristics, and Outcomes of Leukemoid Reaction in 77 Patients*
Demographic and Clinical CharacteristicsResult
  • NSCLC indicates nonsmall cell lung cancer; WBC, white blood cells; bpm, beats per minute.

  • *

    Types of primary tumors: NSCLC (13 patients), sarcoma (11 patients), unknown primary (6 patients), skin (6 patients), pancreas (4 patients), bladder (4 patients), breast (3 patients), gallbladder or bile duct (3 patients), laryngeal/oropharyngeal (3 patients), kidney (2 patients), peritoneal mesothelioma (2 patients), endometrial (2 patients), gastroesophageal junction (2 patients), thyroid (2 patients), bone (1 patient), penile (1 patient), peripheral nerve sheath (1 patient), maxillary sinus (1 patient), tongue (1 patient), stomach (1 patient), liver (1 patient), cervical (1 patient), and ovarian (1 patient).

  • Black (13 patients), Hispanic (8 patients), Middle Eastern (2 patients), and Asian (2 patients).

  • Data were based on 71 patients; 6 were discharged to hospice and were lost to follow-up.

Ethnicity, no. (%) 
 White52 (68)
 Other25 (32)
Median age (range), y56.3 (17-81)
Sex, male/female42 (54.5)/35 (45.5)
Tumor involving lung41 (53)
Tumor metastases60 (78)
NSCLC as sole tumor diagnosis, no. (%)5 (6)
Mean WBC at presentation (range), K/μL53.0 (40.6-115.1)
 Neutrophils >65%, no. (%)74 (96)
 Lymphocytes >40%, no. (%)1 (1)
 Monocytes >10%, no. (%)1 (1)
 Basophils >1%, no. (%)2 (3)
 Eosinophils >5%, no. (%)3 (4)
Serum albumin (n=72) 
 Mean level (range), g/dL3.0 (1.2-4.6)
 <2.0 g/dL, no. (%)4 (6)
Temperature (n=76), no. (%) 
 >38.0°C1 (1)
 >36.0°C and ≤38.0°C65 (86)
 ≤36.0°C10 (13)
Heart rate (n=76), no. (%) 
 ≥100 bpm40 (53)
 <100 bpm36 (47)
Mean arterial pressure (n=76), no. (%) 
 ≥65 mm Hg71 (93)
 <65 mm Hg5 (7)
Respiratory rate (n=76), no. (%) 
 ≤20 breaths/min67 (88)
 >20 breaths/min9 (12)
 Died ≤1 wk12 (17)
 Died 1 ≤4 wk18 (25)
 Died 4 ≤12 wk24 (34)
 Survived >1 y8 (11)
 Lost to follow-up2 (3)

However, the short-term prognosis of the patients with a PLR was poor (Table 2). Six (8%) patients were discharged to hospice and were lost to follow-up. Of the 71 patients for whom data were available, 54 (76%) patients died within 12 weeks, and 67 (89%) patients died within 1 year of their initial extreme leukocyte count. Of the 8 (10%) patients who survived >1 year after their initial presentation, all had resolved leukocytosis after aggressive chemotherapeutic or surgical interventions.


To our knowledge, this is the largest study of patients with solid tumors with extreme leukocytosis presented to date. The majority of cases (90%) had a secondary cause. Only a minority (15%) of the 758 solid tumor patients had an infectious etiology, predominately pneumonia. Nearly 25% of patients had multiple sources of infection.

Seventy-seven (10%) of the 758 solid tumor patients with extreme leukocytosis had a PLR. We likely underrepresented the contribution of PLR to a patient's extreme leukocytosis in that our mutually exclusive diagnostic definitions did not account for the possibility of multiple etiologies. For example, patients may have had a baseline tumor-associated leukocytosis in the range of 20,000/μL to 40,000/μL that was then increased into the extreme range (WBC >40,000/μL) by, for example, a subsequent infection, and therefore would have been defined as having an infectious etiology. A variety of underlying tumors were encountered (Table 2), similar to previous reports. Also similar to other studies, these patients were chronically but not acutely ill, had a neutrophilic predominance (perhaps because neutrophils are the most responsive to a wider variety of cytokines), and had large tumor burdens.1, 7 Previous case series described PLRs only in patients with advanced stage solid tumors,7, 8 although there are isolated case reports of occult metastases to the bone marrow inducing a leukemoid reaction.5 It has been suggested previously that tumor burden correlates with the degree of leukocytosis.19 The significance of the PLR patients having frequent tumor involvement of the lungs is unclear, although this may be simply a marker of large tumor burden. In addition, the extensive epithelial lining of the lungs may make tumor involvement there more likely to produce G-CSF.14

Similar to previous studies, the clinical outcomes of most of the 77 patients with PLR were poor, with 60 (78%) patients either discharged to hospice or dying within 12 weeks of the first extreme leukocyte count. Exceptions did exist for the subset of these patients (10%) who had resolution of their leukocytosis and survived>1 year after successful treatment of their tumors.

In conclusion, extreme leukocytosis in solid tumor patients is infrequently due to infection. PLR should be included in the differential diagnosis of extreme leukocytosis, especially if the patient is clinically stable and has a large tumor burden. Clinical outcomes are poor unless effective antineoplastic treatment is received.


We thank Dr. Mark Routbort for compiling our patient list.

Conflict of Interest Disclosures

Dr. Kontoyiannis has received research support and honoraria from Schering Plough, Pfizer, Astellas Pharma, Enzon Pharmaceuticals, and Merck.