- Email: [email protected]
Diagnosis and Outcome of 100 Consecutive Patients with Extreme Granulocytic Leukocytosis
Diagnosis and Outcome of 100 Consecutive Patients with Extreme Granulocytic Leukocytosis Mark T. Reding, MD, Jonathan R. Hibbs, MD, Vicki A. Morrison, MD, William R. Swaim, MD, Gregory A. Filice, MD PURPOSE: To determine the clinical features, causes, and prognostic significance of extreme leukocytosis in adults. PATIENTS AND METHODS: Medical records of 100 consecutive patients who presented at the Minneapolis Veterans Affairs Medical Center between March 1993 and January 1994 with more than 25,000 leukocytes/mL blood and with more than 50% granulocytes were reviewed. Demographic, clinical, and outcome information was recorded, and a cause of extreme leukocytosis was sought in each case. RESULTS: Extreme leukocytosis was attributed to infection in 48 cases, advanced malignancy in 13 cases, hemorrhage in 9 cases, glucocorticoids in 8 cases, and other causes in 22 cases. Four patients had previously diagnosed conditions resulting in chronic leukocytosis. Higher leukocyte counts were associated
with malignancy (x2 for trend512.5, P ,0.002). Fever was more common in patients with infection (weighted rate ratio53.7, 95% Confidence interval [CI]52.2 to 6.2). Mortality was high overall (31%), and was greater in patients with noninfectious diagnoses compared with infected patients, an association which persisted after stratification by leukocyte count (weighted rate ratio52.5, 95% CI51.2 to 4.9). CONCLUSION: Clinicians should be aware that extreme leukocytosis with a predominance of granulocytes is associated with infection in only 48% of cases. The presence of fever increases the likelihood that infection is the cause. Mortality is high, particularly in patients without infection. Am J Med. 1998;104:12–16. q1998 by Excerpta Medica, Inc.
E
PATIENTS AND METHODS
xtreme leukocytosis is frequently alarming to clinicians and often leads to extensive evaluation and empirical treatment. Although extreme leukocytosis may occur in a variety of infectious, malignant, and inflammatory conditions and also in association with certain medications, little has been written about this phenomenon in the last several decades. The clinical significance of extreme leukocytosis in a large series of patients has not been well described. The term ‘‘leukemoid reaction’’ has been used to describe patients with dramatically elevated leukocyte counts or leukocytosis with a marked left shift that simulates leukemia, but specific definitions of this term vary widely (1–5). The absence of a widely accepted definition hampers understanding of this topic. We reviewed the clinical course of 100 patients with extreme leukocytosis that was predominantly granulocytic. Our objectives were to determine the causes and prognostic significance of extreme leukocytosis in adults.
From the Infectious Disease Section (JRH, VAM, GAF); and Medical Service, Veterans Affairs Medical Center (MTR, WRS); and the Infectious Diseases Division (JRH, VAM, GAF); and Department of Medicine, University of Minnesota (MTR, WRS); and the Department of Laboratory Medicine and Pathology, Veterans Affairs Medical Center and University of Minnesota, Minneapolis, Minnesota (WRS). Presented in part at the Midwestern Section, American Federation for Clinical Research regional meeting, Chicago, Illinois, September 18 to 20, 1996. Requests for reprints should be addressed to Gregory A. Filice, MD, Infectious Disease Section (111F), Veterans Affairs Medical Center, Minneapolis, Minnesota 55417. Manuscript submitted February 11, 1997 and accepted in revised form August 9, 1997. 12
q1998 by Excerpta Medica, Inc. All rights reserved.
We reviewed medical records of 135 consecutive patients who presented to the Minneapolis Veterans Affairs Medical Center, an acute care 523 bed teaching hospital, from March 1993 to January 1994. Cases were identified by review of the hematology laboratory logs. Extreme leukocytosis was defined as a leukocyte count of greater than 25,000/mL on at least 1 occasion with granulocytes accounting for more than 50% of the leukocytes. In our hematology laboratory, a leukocyte count of more than 25,000/mL blood is considered a critical value and is the value at which clinicians must be directly notified of the leukocyte count. Medical records were reviewed by a single investigator with a standardized case report form. Demographic information, significant underlying medical conditions, and details of the clinical presentation, course, and outcome were recorded. From these data and the impressions of the physician(s) caring for each patient, a cause of extreme leukocytosis was identified in each case. Leukocytosis was attributed to infection when there was a positive culture from a normally sterile body site and/or an improvement in the clinical condition and resolution of leukocytosis in response to antimicrobial therapy. Fever was defined as an oral temperature of greater than 388C (100.58F) within 48 hours of the peak leukocyte count. Rectal or axillary temperatures were adjusted to corresponding oral temperatures by subtraction or addition of 0.38C, respectively (6). Pneumonia was defined as the presence of cough, dyspnea, and/or purulent sputum, and a new infiltrate on chest x-ray. Cough, dyspnea, 0002-9343/98/$19.00 PII S0002-9343(97)00273-8
Diagnosis and Outcome of Patients with Granulocytic Leukocytosis/Reding et al
and/or purulent sputum without an infiltrate on chest x-ray was defined as bronchitis. Urinary tract infection (UTI) was defined as the presence of pyuria (more than seven white blood cells/high power field) and a positive urine culture. Systemic inflammatory response syndrome (SIRS), sepsis, severe sepsis, and septic shock were defined using previously established criteria (7). Infections were considered nosocomial when their onset occurred more than 48 hours after hospital admission. In patients with malignancies, leukocytosis was attributed to the malignancy when there was no additional clinical evidence of infection or other likely cause for leukocytosis. Analysis of mortality was based on survival until hospital discharge or, for outpatients, until the next clinic visit. Relative risk (RR), 95% confidence intervals (95% CI), x2 for trend, and Mantel-Haenszel weighted rate ratios were calculated with Epi Info version 6.0 (Centers for Disease Control and Prevention, Atlanta, Georgia).
RESULTS From March 1993 through January 1994, 135 consecutive cases of leukocytosis with greater than 25,000/mL were identified. Among these, 100 were in patients with more than 50% granulocytes. Thirteen of these 100 patients had 50% to 75% granulocytes, 46 had 75.1% to 90% granulocytes, and 41 had greater than 90% granulocytes. The remaining 35 patients had chronic lymphocytic leukemia and more than 50% lymphocytes, and were excluded by the case definition. Of the 100 patients with more than 50% granulocytes, 98 were men, 99 were Caucasian, and the age range was 33 to 101 (median 68) years. These characteristics were similar to those of all patients seen at our hospital. The leukocyte counts ranged from 25,100 to 183,400 (median 31,000)/mL blood. Fortythree patients had leukocyte counts in the 25,000 to 30,000/mL range, 25 had leukocyte counts in the 30,100 to 35,000/mL range, and 12 had leukocyte counts from 35,100 to 40,000/mL. The remaining 20 patients had leukocyte counts scattered throughout the 40,100 to 183,400/mL range. Four patients had conditions resulting in chronic extreme leukocytosis (1 case each of polycythemia vera, myeloid metaplasia, chronic myeloid leukemia, and absence of the spleen). The remaining 96 patients had conditions resulting in acute extreme leukocytosis. During the study period there were 14,142 hospital discharges; the incidence was 7 cases of extreme leukocytosis per 1000 discharges. A condition that led to extreme leukocytosis was identified in 98 of the 100 cases (Table 1). Extreme leukocytosis was a consequence of infection in 48 patients, with pneumonia and UTI accounting for 29 (60%) of the 48 infections. Seventeen (35%) of the infections were noso-
comial. All infected patients received antimicrobial therapy. An additional 19 patients received antimicrobial therapy and were found not to have extreme leukocytosis as a consequence of infection. Extreme leukocytosis was attributed to malignancy in 15 of the 100 patients studied. All six patients with solid tumors had metastatic disease which was readily apparent on presentation. Of the nine patients with hematologic malignancies, only two were newly diagnosed at the time extreme leukocytosis was detected. In 35 of the remaining 37 patients, other less common causes of extreme leukocytosis were identified. Twelve patients were receiving medications known to cause leu-
Table 1. Conditions Associated with Extreme Leukocytosis
Infection Pneumonia Urinary tract infection Sepsis/severe sepsis/septic shock Abscess* Bronchitis Cellulitis Diabetic foot ulcer Bacteremia following ERCP Malignancy Metastatic carcinoma† Non-Hodgkin’s lymphoma Chronic myeloid leukemia‡ Acute myeloid leukemia§ Polycythemia vera Myeloid metaplasia Other Hemorrhage\ Glucocorticoid therapy G-CSF therapy Cardiac arrest SIRS Hepatorenal syndrome Miscellaneous¶ Unknown Total
Number of Cases
Number of Deaths
48 17 12 9 3 3 2 1 1 15 6 3 3 1 1 1 37 9 8 4 4 2 2 6 2 100
8 (17%) 1 2 2 2 1 0 0 0 8 (53%) 6 1 0 1 0 0 15 (41%) 4 1 0 3 2 2 1 2 31 (31%)
* 1 Intra-abdominal, 1 perirectal, 1 retroperitoneal. † 3 Non–small-cell lung cancer, 2 bladder cancer (transitional cell carcinoma), 1 poorly differentiated carcinoma of unknown primary. ‡ 2 Chronic phase, 1 accelerated phase. § Transformed from chronic myelomonocytic leukemia. \ 7 gastrointestinal, 2 complications of angiography. ¶ 1 Acute renal failure, 1 diabetic ketoacidosis, 1 pancreatic phlegmon (not infected), 1 abdominal pain after barium enema, 1 myocardial infarction (uncomplicated, postoperative), 1 splenectomy. ERCP 5 endoscopic retrograde cholangiopancreatography; G-CSF 5 granulocyte-colony stimulating factor; SIRS 5 Systemic inflammatory response syndrome.
January 1998
THE AMERICAN JOURNAL OF MEDICINEt
Volume 104 13
Diagnosis and Outcome of Patients with Granulocytic Leukocytosis/Reding et al
Table 2. Drugs and Extreme Leukocytosis Patient Number 1 2 3 4 5 6 7 8 9 10 11 12
Indication
Drug/Dose
Duration Before Leukocytosis
Peak Leukocyte Count (31000)/mL
COPD COPD COPD Amiodarone toxicity Adrenal insufficiency Adrenal insufficiency TTP Lymphoma Leukopenia Leukopenia Leukopenia Leukopenia
Methylprednisolone 125 mg q6h Methylprednisolone 60 mg q6h Methylprednisolone 125 mg q6h Methylprednisolone 125 mg qd Hydrocortisone 100 mg q8h Hydrocortisone 100 mg qd Prednisone 100 mg qd Predisone 220 mg qd G-CSF 5 mg/kg/d G-CSF 5 mg/kg/d G-CSF 5 mg/kg/d G-CSF 5 mg/kg/d
2d 4d 2d 6d 2d 3d 12d 2d 9d 12d 12d 21d
27.6 33.9 26.3 30.7 30.9 30.7 29.8 36.3 25.5 26.9 60.8 60.4
COPD 5 chronic obstructive pulmonary disease; TTP 5 thrombotic thrombocytopenic purpura; G-CSF 5 granulocyte-colony stimulating factor.
kocytosis, including eight patients treated with systemic glucocorticoids and four patients who received granulocyte-colony stimulating factor following chemotherapy (Table 2). Nine patients had extreme leukocytosis secondary to significant blood loss, including seven with gastrointestinal hemorrhage, and two who had hemorrhage as a complication of angiography. A specific cause for extreme leukocytosis was not apparent in two cases. Both patients died, and autopsies were not performed. The presence or absence of fever was predictive of the etiology of extreme leukocytosis. Eighty percent of patients with extreme leukocytosis and fever had infection. In contrast, only 22% of patients with extreme leukocytosis and no fever had infection (weighted rate ratio 5 3.7, 95% CI 5 2.2 to 6.2). The degree of leukocytosis could be used to predict etiology (Figure 1). Malignancy was an uncommon cause of extreme leukocytosis in patients with leukocyte counts less than 45,000/mL, accounting for only 10% of cases. In contrast, in patients with leukocyte counts greater than 45,000/mL, malignancy was the most common cause, ac-
counting for 44% of cases (x2 for trend 5 12.5, P ,0.002). The leukocyte differential count was not predictive of the condition associated with extreme leukocytosis. Specifically, a higher percentage of neutrophil band forms was not associated with any particular diagnostic category (ie, infection). Similarly, a higher total percentage of granulocytes was not predictive of any particular diagnostic category (data not shown). Thirty-one patients died. Mortality was higher among patients with noninfectious diagnoses compared with infected patients (Table 3). This association persisted after stratification by leukocyte count (weighted rate ratio 5 2.5, 95% CI 5 1.2 to 4.9). There were no significant associations between leukocyte count and mortality within diagnostic categories.
DISCUSSION Our data represent the largest series of patients in decades in which the clinical significance of extreme leukocytosis
Figure 1. Diagnostic category and degree of leukocytosis.
14
January 1998
THE AMERICAN JOURNAL OF MEDICINEt Volume 104
Diagnosis and Outcome of Patients with Granulocytic Leukocytosis/Reding et al
Table 3. Diagnosis, Leukocyte Count, and Mortality
Diagnosis
Leukocytes/mL
Infection
25,000–35,000 35,001–45,000 .45,000 subtotal 25,000–35,000 35,001–45,000 .45,000 subtotal 25,000–35,000 35,001–45,000 .45,000 subtotal
Malignancy
Other
Total
Number of Cases
Number of Deaths (%)
34 9 5 48 6 2 7 15 28 5 4 37 100
4 (12) 2 (22) 2 (40) 8 (17) 3 (50) 1 (50) 4 (57) 8 (53) 11 (39) 2 (40) 2 (50) 15 (41) 31 (31)
has been examined in a comprehensive fashion. This analysis provides clinicians with practical information to guide the diagnostic and therapeutic management of patients with extreme leukocytosis. We found that extreme leukocytosis was associated with a wide variety of medical conditions. While infection is often high on the list of differential diagnoses in patients with leukocytosis of any magnitude, infection was present in slightly less than half of the patients in our series. Of the noninfectious causes that accounted for just over half of our cases, malignancy was the most common associated condition. Mild to extreme elevations in the leukocyte count occur in a patients with a variety of tumors and are present in the absence of infection in approximately 25% of cancer patients (8). The other noninfectious conditions in our series have been previously described in association with leukocytosis, although not necessarily extreme leukocytosis (7,9,10). In view of the many noninfectious causes of extreme leukocytosis, this laboratory finding does not in itself constitute an indication for empirical antimicrobial therapy in an otherwise stable and afebrile patient. The presence of fever in combination with extreme leukocytosis was strongly associated with underlying infection. While other investigators have previously examined the relationships between fever, elevated leukocyte counts, and infection, the clinical significance of fever has not been documented in a series of patients with extreme leukocytosis. In patients who have undergone cardiac surgery, fever and leukocytosis have been found to be insensitive and nonspecific signs of infection (11,12). In contrast, the presence of fever and leukocytosis greater than 15,000/mL is predictive of an increased risk of occult bacterial infection in both children (13–15) and adults (16). Our findings complement these prior observations
and suggest that fever is a reliable predictor of infection in adult patients with extreme leukocytosis. The degree of leukocytosis also reflects the underlying etiology in patients with extreme leukocytosis. Malignancy was significantly less likely to be the cause of extreme leukocytosis in our patients with leukocyte counts less than 45,000/mL than in those with higher leukocyte counts. The diagnosis was readily apparent in all cases with malignancy and did not require extensive diagnostic testing. Thus, in patients with extreme leukocytosis, an extensive evaluation for an occult malignancy is likely to be of low yield, particularly in patients with leukocyte counts less than 45,000/mL. Mortality was high in patients with extreme leukocytosis, particularly those with noninfectious diagnoses. One third of the 100 patients in our series did not survive until hospital discharge during the admission in which extreme leukocytosis occurred. Although previous data on patients with extreme leukocytosis are lacking, investigators have demonstrated an association between leukocytosis and severity of coronary artery disease, (17,18) severity of depression, (19) and degree of injury in trauma patients (20). Others have observed higher mortality in patients with malignancy, (17) myocardial infarction, (21) head trauma, (22) subarachnoid hemorrhage, (23) and alcoholic hepatitis (5). Our data are consistent with the general observation that a higher degree of leukocytosis reflects greater severity of illness and poorer prognosis.
ACKNOWLEDGEMENTS We are grateful to the staff of the Hematology Laboratory of the Minneapolis Veterans Affairs Medical Center for finding our cases.
REFERENCES 1. Hilts SV, Shaw CC. Leukemoid blood reactions. NEJM. 1953;249: 434 – 438. 2. Ward PCJ. The myeloid leukocytoses. Postgrad Med. 1980;67:219 – 229. 3. Schmid C, Frisch B, Beham A, et al. Comparison of bone marrow histology in early chronic granulocytic leukemia and in leukemoid reaction. Eur J Haematol. 1990;44:154 –158. 4. Cvitkovic E, Bachouchi M, Boussen H, et al. Leukemoid reaction, bone marrow invasion, fever of unknown origin, and metastatic pattern in the natural history of advanced undifferentiated carcinoma of nasopharyngeal type: A review of 255 consecutive cases. J Clin Oncol. 1993;11:2434 –2442. 5. Mitchell RG, Michael M, Sandidge S. High mortality among patients with the leukemoid reaction and alcoholic hepatitis. South Med J. 1991;84:281–282. 6. Cranston WI, Gerbrandy J, Snell ES. Oral, rectal, and oesophageal temperatures and some factors affecting them in man. J Physiol. 1954;126:347–358. 7. Bone RC, Balk RA, Cerra FB, et al. Definitions for sepsis and organ failure and guidelines for the use of innovative therapies in sepsis. Chest. 1992;101:1644 –1655. January 1998
THE AMERICAN JOURNAL OF MEDICINEt
Volume 104 15
Diagnosis and Outcome of Patients with Granulocytic Leukocytosis/Reding et al 8. Johnson RA, Roodman GD. Hematologic manifestations of malignancy. Dis Mon. 1989;35:721–768. 9. Gallin JI. Quantitative and qualitative disorders of phagocytes. In Isselbacher KJ, Braunwald E, Wilson JD, et al., eds. Harrison’s Principles of Internal Medicine. 13th ed. New York:McGraw-Hill; 1994: 329 –337. 10. Dale DC. Neutrophilia. In Williams WJ, Buetler E, Erslev AJ, Lichtman MA, eds. Hematology. 4th ed. New York:McGraw-Hill; 1990: 816 – 820. 11. Bell DM, Goldman DA, Hopkins CC, et al. Unreliability of fever and leukocytosis in the diagnosis of infection after cardiac valve surgery. J Thorac Cardiovasc Surg. 1978;75:87–90. 12. Miholic J, Hiertz H, Hudec M, et al. Fever, leukocytosis, and infection after open heart surgery. A log-linear regression analysis of 115 cases. Thorac Cardiovasc Surg. 1984;32:45– 48. 13. Werman HA, Brown CG. White blood cell count and differential count. Emerg Med Clin North Am. 1986;4:41–58. 14. Singer JI, Buchino JJ, Chabali R. Selected laboratory in pediatric emergency care. Emerg Med Clin North Am. 1986;4:377–396. 15. Talan DA. Infectious disease issues in the emergency department. Clin Infect Dis. 1996;23:1–14.
16
January 1998
THE AMERICAN JOURNAL OF MEDICINEt Volume 104
16. Mellors JW, Horwitz RI, Harvey MR, Horwitz SM. A simple index to identify occult bacterial infection in adults with acute unexplained fever. Arch Intern Med. 1987;147:666 – 671. 17. Grimm RH, Neaton JD, Ludwig W. Prognostic importance of the white blood cell count for coronary, cancer, and all-cause mortality. JAMA. 1985;254:1932–1937. 18. Zalokar JB, Richard JL, Claude JR. Leukocyte count, smoking, and myocardial infarction. NEJM. 1981;304:465– 468. 19. Maes M, Van Der Planken M, Stevens WJ, et al. Leukocytosis, monocytosis, and neutrophilia: Hallmarks of severe depression. J Psychiat Res. 1992;26:125–134. 20. Morell V, Lundgren E, Gillott A. Predicting severity of trauma by admission white blood cell count, serum potassium level, and arterial pH. South Med J. 1993;86:658 – 659. 21. Modan B, Schor S, Shani M. Acute myocardial infarction: Prognostic value of white blood cell count and glucose level. JAMA. 1975; 233:266 –267. 22. Keskil S, Baykaner MK, Cevikev N, Aykol S. Head trauma and leukocytosis. Acta Neurochir. 1994;131:211–214. 23. Parkinson D, Stephensen S. Leukocytosis and subarachnoid hemorrhage. Surg Neurol. 1984;21:132–134.
with malignancy (x2 for trend512.5, P ,0.002). Fever was more common in patients with infection (weighted rate ratio53.7, 95% Confidence interval [CI]52.2 to 6.2). Mortality was high overall (31%), and was greater in patients with noninfectious diagnoses compared with infected patients, an association which persisted after stratification by leukocyte count (weighted rate ratio52.5, 95% CI51.2 to 4.9). CONCLUSION: Clinicians should be aware that extreme leukocytosis with a predominance of granulocytes is associated with infection in only 48% of cases. The presence of fever increases the likelihood that infection is the cause. Mortality is high, particularly in patients without infection. Am J Med. 1998;104:12–16. q1998 by Excerpta Medica, Inc.
E
PATIENTS AND METHODS
xtreme leukocytosis is frequently alarming to clinicians and often leads to extensive evaluation and empirical treatment. Although extreme leukocytosis may occur in a variety of infectious, malignant, and inflammatory conditions and also in association with certain medications, little has been written about this phenomenon in the last several decades. The clinical significance of extreme leukocytosis in a large series of patients has not been well described. The term ‘‘leukemoid reaction’’ has been used to describe patients with dramatically elevated leukocyte counts or leukocytosis with a marked left shift that simulates leukemia, but specific definitions of this term vary widely (1–5). The absence of a widely accepted definition hampers understanding of this topic. We reviewed the clinical course of 100 patients with extreme leukocytosis that was predominantly granulocytic. Our objectives were to determine the causes and prognostic significance of extreme leukocytosis in adults.
From the Infectious Disease Section (JRH, VAM, GAF); and Medical Service, Veterans Affairs Medical Center (MTR, WRS); and the Infectious Diseases Division (JRH, VAM, GAF); and Department of Medicine, University of Minnesota (MTR, WRS); and the Department of Laboratory Medicine and Pathology, Veterans Affairs Medical Center and University of Minnesota, Minneapolis, Minnesota (WRS). Presented in part at the Midwestern Section, American Federation for Clinical Research regional meeting, Chicago, Illinois, September 18 to 20, 1996. Requests for reprints should be addressed to Gregory A. Filice, MD, Infectious Disease Section (111F), Veterans Affairs Medical Center, Minneapolis, Minnesota 55417. Manuscript submitted February 11, 1997 and accepted in revised form August 9, 1997. 12
q1998 by Excerpta Medica, Inc. All rights reserved.
We reviewed medical records of 135 consecutive patients who presented to the Minneapolis Veterans Affairs Medical Center, an acute care 523 bed teaching hospital, from March 1993 to January 1994. Cases were identified by review of the hematology laboratory logs. Extreme leukocytosis was defined as a leukocyte count of greater than 25,000/mL on at least 1 occasion with granulocytes accounting for more than 50% of the leukocytes. In our hematology laboratory, a leukocyte count of more than 25,000/mL blood is considered a critical value and is the value at which clinicians must be directly notified of the leukocyte count. Medical records were reviewed by a single investigator with a standardized case report form. Demographic information, significant underlying medical conditions, and details of the clinical presentation, course, and outcome were recorded. From these data and the impressions of the physician(s) caring for each patient, a cause of extreme leukocytosis was identified in each case. Leukocytosis was attributed to infection when there was a positive culture from a normally sterile body site and/or an improvement in the clinical condition and resolution of leukocytosis in response to antimicrobial therapy. Fever was defined as an oral temperature of greater than 388C (100.58F) within 48 hours of the peak leukocyte count. Rectal or axillary temperatures were adjusted to corresponding oral temperatures by subtraction or addition of 0.38C, respectively (6). Pneumonia was defined as the presence of cough, dyspnea, and/or purulent sputum, and a new infiltrate on chest x-ray. Cough, dyspnea, 0002-9343/98/$19.00 PII S0002-9343(97)00273-8
Diagnosis and Outcome of Patients with Granulocytic Leukocytosis/Reding et al
and/or purulent sputum without an infiltrate on chest x-ray was defined as bronchitis. Urinary tract infection (UTI) was defined as the presence of pyuria (more than seven white blood cells/high power field) and a positive urine culture. Systemic inflammatory response syndrome (SIRS), sepsis, severe sepsis, and septic shock were defined using previously established criteria (7). Infections were considered nosocomial when their onset occurred more than 48 hours after hospital admission. In patients with malignancies, leukocytosis was attributed to the malignancy when there was no additional clinical evidence of infection or other likely cause for leukocytosis. Analysis of mortality was based on survival until hospital discharge or, for outpatients, until the next clinic visit. Relative risk (RR), 95% confidence intervals (95% CI), x2 for trend, and Mantel-Haenszel weighted rate ratios were calculated with Epi Info version 6.0 (Centers for Disease Control and Prevention, Atlanta, Georgia).
RESULTS From March 1993 through January 1994, 135 consecutive cases of leukocytosis with greater than 25,000/mL were identified. Among these, 100 were in patients with more than 50% granulocytes. Thirteen of these 100 patients had 50% to 75% granulocytes, 46 had 75.1% to 90% granulocytes, and 41 had greater than 90% granulocytes. The remaining 35 patients had chronic lymphocytic leukemia and more than 50% lymphocytes, and were excluded by the case definition. Of the 100 patients with more than 50% granulocytes, 98 were men, 99 were Caucasian, and the age range was 33 to 101 (median 68) years. These characteristics were similar to those of all patients seen at our hospital. The leukocyte counts ranged from 25,100 to 183,400 (median 31,000)/mL blood. Fortythree patients had leukocyte counts in the 25,000 to 30,000/mL range, 25 had leukocyte counts in the 30,100 to 35,000/mL range, and 12 had leukocyte counts from 35,100 to 40,000/mL. The remaining 20 patients had leukocyte counts scattered throughout the 40,100 to 183,400/mL range. Four patients had conditions resulting in chronic extreme leukocytosis (1 case each of polycythemia vera, myeloid metaplasia, chronic myeloid leukemia, and absence of the spleen). The remaining 96 patients had conditions resulting in acute extreme leukocytosis. During the study period there were 14,142 hospital discharges; the incidence was 7 cases of extreme leukocytosis per 1000 discharges. A condition that led to extreme leukocytosis was identified in 98 of the 100 cases (Table 1). Extreme leukocytosis was a consequence of infection in 48 patients, with pneumonia and UTI accounting for 29 (60%) of the 48 infections. Seventeen (35%) of the infections were noso-
comial. All infected patients received antimicrobial therapy. An additional 19 patients received antimicrobial therapy and were found not to have extreme leukocytosis as a consequence of infection. Extreme leukocytosis was attributed to malignancy in 15 of the 100 patients studied. All six patients with solid tumors had metastatic disease which was readily apparent on presentation. Of the nine patients with hematologic malignancies, only two were newly diagnosed at the time extreme leukocytosis was detected. In 35 of the remaining 37 patients, other less common causes of extreme leukocytosis were identified. Twelve patients were receiving medications known to cause leu-
Table 1. Conditions Associated with Extreme Leukocytosis
Infection Pneumonia Urinary tract infection Sepsis/severe sepsis/septic shock Abscess* Bronchitis Cellulitis Diabetic foot ulcer Bacteremia following ERCP Malignancy Metastatic carcinoma† Non-Hodgkin’s lymphoma Chronic myeloid leukemia‡ Acute myeloid leukemia§ Polycythemia vera Myeloid metaplasia Other Hemorrhage\ Glucocorticoid therapy G-CSF therapy Cardiac arrest SIRS Hepatorenal syndrome Miscellaneous¶ Unknown Total
Number of Cases
Number of Deaths
48 17 12 9 3 3 2 1 1 15 6 3 3 1 1 1 37 9 8 4 4 2 2 6 2 100
8 (17%) 1 2 2 2 1 0 0 0 8 (53%) 6 1 0 1 0 0 15 (41%) 4 1 0 3 2 2 1 2 31 (31%)
* 1 Intra-abdominal, 1 perirectal, 1 retroperitoneal. † 3 Non–small-cell lung cancer, 2 bladder cancer (transitional cell carcinoma), 1 poorly differentiated carcinoma of unknown primary. ‡ 2 Chronic phase, 1 accelerated phase. § Transformed from chronic myelomonocytic leukemia. \ 7 gastrointestinal, 2 complications of angiography. ¶ 1 Acute renal failure, 1 diabetic ketoacidosis, 1 pancreatic phlegmon (not infected), 1 abdominal pain after barium enema, 1 myocardial infarction (uncomplicated, postoperative), 1 splenectomy. ERCP 5 endoscopic retrograde cholangiopancreatography; G-CSF 5 granulocyte-colony stimulating factor; SIRS 5 Systemic inflammatory response syndrome.
January 1998
THE AMERICAN JOURNAL OF MEDICINEt
Volume 104 13
Diagnosis and Outcome of Patients with Granulocytic Leukocytosis/Reding et al
Table 2. Drugs and Extreme Leukocytosis Patient Number 1 2 3 4 5 6 7 8 9 10 11 12
Indication
Drug/Dose
Duration Before Leukocytosis
Peak Leukocyte Count (31000)/mL
COPD COPD COPD Amiodarone toxicity Adrenal insufficiency Adrenal insufficiency TTP Lymphoma Leukopenia Leukopenia Leukopenia Leukopenia
Methylprednisolone 125 mg q6h Methylprednisolone 60 mg q6h Methylprednisolone 125 mg q6h Methylprednisolone 125 mg qd Hydrocortisone 100 mg q8h Hydrocortisone 100 mg qd Prednisone 100 mg qd Predisone 220 mg qd G-CSF 5 mg/kg/d G-CSF 5 mg/kg/d G-CSF 5 mg/kg/d G-CSF 5 mg/kg/d
2d 4d 2d 6d 2d 3d 12d 2d 9d 12d 12d 21d
27.6 33.9 26.3 30.7 30.9 30.7 29.8 36.3 25.5 26.9 60.8 60.4
COPD 5 chronic obstructive pulmonary disease; TTP 5 thrombotic thrombocytopenic purpura; G-CSF 5 granulocyte-colony stimulating factor.
kocytosis, including eight patients treated with systemic glucocorticoids and four patients who received granulocyte-colony stimulating factor following chemotherapy (Table 2). Nine patients had extreme leukocytosis secondary to significant blood loss, including seven with gastrointestinal hemorrhage, and two who had hemorrhage as a complication of angiography. A specific cause for extreme leukocytosis was not apparent in two cases. Both patients died, and autopsies were not performed. The presence or absence of fever was predictive of the etiology of extreme leukocytosis. Eighty percent of patients with extreme leukocytosis and fever had infection. In contrast, only 22% of patients with extreme leukocytosis and no fever had infection (weighted rate ratio 5 3.7, 95% CI 5 2.2 to 6.2). The degree of leukocytosis could be used to predict etiology (Figure 1). Malignancy was an uncommon cause of extreme leukocytosis in patients with leukocyte counts less than 45,000/mL, accounting for only 10% of cases. In contrast, in patients with leukocyte counts greater than 45,000/mL, malignancy was the most common cause, ac-
counting for 44% of cases (x2 for trend 5 12.5, P ,0.002). The leukocyte differential count was not predictive of the condition associated with extreme leukocytosis. Specifically, a higher percentage of neutrophil band forms was not associated with any particular diagnostic category (ie, infection). Similarly, a higher total percentage of granulocytes was not predictive of any particular diagnostic category (data not shown). Thirty-one patients died. Mortality was higher among patients with noninfectious diagnoses compared with infected patients (Table 3). This association persisted after stratification by leukocyte count (weighted rate ratio 5 2.5, 95% CI 5 1.2 to 4.9). There were no significant associations between leukocyte count and mortality within diagnostic categories.
DISCUSSION Our data represent the largest series of patients in decades in which the clinical significance of extreme leukocytosis
Figure 1. Diagnostic category and degree of leukocytosis.
14
January 1998
THE AMERICAN JOURNAL OF MEDICINEt Volume 104
Diagnosis and Outcome of Patients with Granulocytic Leukocytosis/Reding et al
Table 3. Diagnosis, Leukocyte Count, and Mortality
Diagnosis
Leukocytes/mL
Infection
25,000–35,000 35,001–45,000 .45,000 subtotal 25,000–35,000 35,001–45,000 .45,000 subtotal 25,000–35,000 35,001–45,000 .45,000 subtotal
Malignancy
Other
Total
Number of Cases
Number of Deaths (%)
34 9 5 48 6 2 7 15 28 5 4 37 100
4 (12) 2 (22) 2 (40) 8 (17) 3 (50) 1 (50) 4 (57) 8 (53) 11 (39) 2 (40) 2 (50) 15 (41) 31 (31)
has been examined in a comprehensive fashion. This analysis provides clinicians with practical information to guide the diagnostic and therapeutic management of patients with extreme leukocytosis. We found that extreme leukocytosis was associated with a wide variety of medical conditions. While infection is often high on the list of differential diagnoses in patients with leukocytosis of any magnitude, infection was present in slightly less than half of the patients in our series. Of the noninfectious causes that accounted for just over half of our cases, malignancy was the most common associated condition. Mild to extreme elevations in the leukocyte count occur in a patients with a variety of tumors and are present in the absence of infection in approximately 25% of cancer patients (8). The other noninfectious conditions in our series have been previously described in association with leukocytosis, although not necessarily extreme leukocytosis (7,9,10). In view of the many noninfectious causes of extreme leukocytosis, this laboratory finding does not in itself constitute an indication for empirical antimicrobial therapy in an otherwise stable and afebrile patient. The presence of fever in combination with extreme leukocytosis was strongly associated with underlying infection. While other investigators have previously examined the relationships between fever, elevated leukocyte counts, and infection, the clinical significance of fever has not been documented in a series of patients with extreme leukocytosis. In patients who have undergone cardiac surgery, fever and leukocytosis have been found to be insensitive and nonspecific signs of infection (11,12). In contrast, the presence of fever and leukocytosis greater than 15,000/mL is predictive of an increased risk of occult bacterial infection in both children (13–15) and adults (16). Our findings complement these prior observations
and suggest that fever is a reliable predictor of infection in adult patients with extreme leukocytosis. The degree of leukocytosis also reflects the underlying etiology in patients with extreme leukocytosis. Malignancy was significantly less likely to be the cause of extreme leukocytosis in our patients with leukocyte counts less than 45,000/mL than in those with higher leukocyte counts. The diagnosis was readily apparent in all cases with malignancy and did not require extensive diagnostic testing. Thus, in patients with extreme leukocytosis, an extensive evaluation for an occult malignancy is likely to be of low yield, particularly in patients with leukocyte counts less than 45,000/mL. Mortality was high in patients with extreme leukocytosis, particularly those with noninfectious diagnoses. One third of the 100 patients in our series did not survive until hospital discharge during the admission in which extreme leukocytosis occurred. Although previous data on patients with extreme leukocytosis are lacking, investigators have demonstrated an association between leukocytosis and severity of coronary artery disease, (17,18) severity of depression, (19) and degree of injury in trauma patients (20). Others have observed higher mortality in patients with malignancy, (17) myocardial infarction, (21) head trauma, (22) subarachnoid hemorrhage, (23) and alcoholic hepatitis (5). Our data are consistent with the general observation that a higher degree of leukocytosis reflects greater severity of illness and poorer prognosis.
ACKNOWLEDGEMENTS We are grateful to the staff of the Hematology Laboratory of the Minneapolis Veterans Affairs Medical Center for finding our cases.
REFERENCES 1. Hilts SV, Shaw CC. Leukemoid blood reactions. NEJM. 1953;249: 434 – 438. 2. Ward PCJ. The myeloid leukocytoses. Postgrad Med. 1980;67:219 – 229. 3. Schmid C, Frisch B, Beham A, et al. Comparison of bone marrow histology in early chronic granulocytic leukemia and in leukemoid reaction. Eur J Haematol. 1990;44:154 –158. 4. Cvitkovic E, Bachouchi M, Boussen H, et al. Leukemoid reaction, bone marrow invasion, fever of unknown origin, and metastatic pattern in the natural history of advanced undifferentiated carcinoma of nasopharyngeal type: A review of 255 consecutive cases. J Clin Oncol. 1993;11:2434 –2442. 5. Mitchell RG, Michael M, Sandidge S. High mortality among patients with the leukemoid reaction and alcoholic hepatitis. South Med J. 1991;84:281–282. 6. Cranston WI, Gerbrandy J, Snell ES. Oral, rectal, and oesophageal temperatures and some factors affecting them in man. J Physiol. 1954;126:347–358. 7. Bone RC, Balk RA, Cerra FB, et al. Definitions for sepsis and organ failure and guidelines for the use of innovative therapies in sepsis. Chest. 1992;101:1644 –1655. January 1998
THE AMERICAN JOURNAL OF MEDICINEt
Volume 104 15
Diagnosis and Outcome of Patients with Granulocytic Leukocytosis/Reding et al 8. Johnson RA, Roodman GD. Hematologic manifestations of malignancy. Dis Mon. 1989;35:721–768. 9. Gallin JI. Quantitative and qualitative disorders of phagocytes. In Isselbacher KJ, Braunwald E, Wilson JD, et al., eds. Harrison’s Principles of Internal Medicine. 13th ed. New York:McGraw-Hill; 1994: 329 –337. 10. Dale DC. Neutrophilia. In Williams WJ, Buetler E, Erslev AJ, Lichtman MA, eds. Hematology. 4th ed. New York:McGraw-Hill; 1990: 816 – 820. 11. Bell DM, Goldman DA, Hopkins CC, et al. Unreliability of fever and leukocytosis in the diagnosis of infection after cardiac valve surgery. J Thorac Cardiovasc Surg. 1978;75:87–90. 12. Miholic J, Hiertz H, Hudec M, et al. Fever, leukocytosis, and infection after open heart surgery. A log-linear regression analysis of 115 cases. Thorac Cardiovasc Surg. 1984;32:45– 48. 13. Werman HA, Brown CG. White blood cell count and differential count. Emerg Med Clin North Am. 1986;4:41–58. 14. Singer JI, Buchino JJ, Chabali R. Selected laboratory in pediatric emergency care. Emerg Med Clin North Am. 1986;4:377–396. 15. Talan DA. Infectious disease issues in the emergency department. Clin Infect Dis. 1996;23:1–14.
16
January 1998
THE AMERICAN JOURNAL OF MEDICINEt Volume 104
16. Mellors JW, Horwitz RI, Harvey MR, Horwitz SM. A simple index to identify occult bacterial infection in adults with acute unexplained fever. Arch Intern Med. 1987;147:666 – 671. 17. Grimm RH, Neaton JD, Ludwig W. Prognostic importance of the white blood cell count for coronary, cancer, and all-cause mortality. JAMA. 1985;254:1932–1937. 18. Zalokar JB, Richard JL, Claude JR. Leukocyte count, smoking, and myocardial infarction. NEJM. 1981;304:465– 468. 19. Maes M, Van Der Planken M, Stevens WJ, et al. Leukocytosis, monocytosis, and neutrophilia: Hallmarks of severe depression. J Psychiat Res. 1992;26:125–134. 20. Morell V, Lundgren E, Gillott A. Predicting severity of trauma by admission white blood cell count, serum potassium level, and arterial pH. South Med J. 1993;86:658 – 659. 21. Modan B, Schor S, Shani M. Acute myocardial infarction: Prognostic value of white blood cell count and glucose level. JAMA. 1975; 233:266 –267. 22. Keskil S, Baykaner MK, Cevikev N, Aykol S. Head trauma and leukocytosis. Acta Neurochir. 1994;131:211–214. 23. Parkinson D, Stephensen S. Leukocytosis and subarachnoid hemorrhage. Surg Neurol. 1984;21:132–134.