Etiology of Thrombocytosis in a General Medicine Population: Analysis of 801 Cases With Emphasis on Infectious Causes

Background The clinical importance of an elevated platelet count is often overlooked, particularly as a diagnostic clue to the presence of an underlying infection. We sought to better describe the relationship between thrombocytosis and inflammatory conditions, with a focus on infectious causes. Methods We retrospectively reviewed 801 sequential cases of thrombocytosis (platelet count > 500 × 109/L) at a tertiary care hospital. Results Essential thrombocythemia was the most common cause of primary thrombocytosis, and these patients were more likely to have extreme (> 800 × 109/L) and prolonged (> 1 month) thrombocytosis. Secondary thrombocytosis was more common than primary, with infectious causes accounting for nearly half the cases. Demographic factors associated with an infectious etiology included inpatient status, quadriplegia/paraplegia, an indwelling prosthesis, dementia and diabetes. Clinical and laboratory characteristics associated with an infectious cause of thrombocytosis included fever, tachycardia, weight loss, hypoalbuminemia, neutrophilia, leukocytosis and anemia. Patients with thrombocytosis secondary to infection had a more rapid normalization of platelet count, but higher risk of dying, than those with secondary, non-infectious causes. Conclusions Infection is a common cause of thrombocytosis and should be considered in patients with comorbidities that increase risk of infection and when clinical and/or laboratory data support an infectious etiology. Thrombocytosis may have prognostic implications as a clinical parameter.


Introduction
The laboratory finding of thrombocytosis, defined as an abnormally elevated platelet count, is not generally recognized as a clinical indicator of infection. However, our clinical experience has suggested that patients with an undrained focus of infection might present with an elevated platelet count, even in the absence of other signs of infection, such as fever or an elevated white blood cell (WBC) count. Furthermore, several observational studies have shown that infection is a common cause of thrombocytosis [1][2][3][4][5][6][7]. More generally, these studies have demonstrated that thrombocytosis is usually reactive in nature, i.e., secondary to an underlying inflammatory condition such as infection, malignancy or tissue damage. By contrast, primary thrombocytosis, or an elevation in platelet count due to a myeloproliferative disorder (MPD), is relatively rare.
Although prior studies have established infection as a cause of thrombocytosis, they lack a detailed description of the underlying infections associated with an elevated platelet count. They also offer little information as to laboratory and clinical signs that may point to an infectious etiology.
Thus, in this study, we describe the etiologies of thrombocytosis in 801 adult patients in a general medicine population, with an emphasis on infectious causes. We report the types of infection associated with thrombocytosis and the causative microorganisms and predisposing conditions, as well as the presence of other clinical or laboratory indicators of infection. We also comment on the relationship between treatment of the underlying infection and clinical outcome, including the potential role of thrombocytosis as a clinical marker of unresolved infection. This observational study adds to the understanding of the association of infection and platelet counts, to aid clinicians in the identification and management of patients with thrombocytosis due to infection.

Clinical setting
The Michael E. DeBakey VA Medical Center (MEDVAMC) is a 520-bed, tertiary care hospital that provides primary medical care for roughly 120,000 veterans and has approximately 15,000 admissions annually. All medical records, including progress notes, nursing notes, consultations, discharge summaries, vital signs, laboratory studies, radiographic and pathologic data, are available electronically.
The protocol was approved by the Institutional Review Board of Baylor College of Medicine and by the Research and Development Committee of the MEDVAMC, and all research activities were conducted at the MEDVAMC / Baylor College of Medicine.

Identification of patients with thrombocytosis
By querying electronic medical records, we retrospectively identified adult patients (≥ 18 years of age) at the MED-VAMC with documented thrombocytosis (platelet count > 500 × 10 9 /L) between March and December 2005. If a patient had multiple episodes of thrombocytosis in the given period, only the episode with the highest platelet count was included. Patients were excluded if the documented platelet count had no associated clinical notes. All patients who met criteria during the study period were included. Overall, the data from 801 patients were analyzed.

Sources for data collection
We recorded demographic data, vital signs and laboratory values, as well as predisposing conditions (such as diabetes and chronic liver disease), admission diagnoses/chief complaints and the presence of indwelling prostheses, including vascular catheters. Data for fever, anemia, tachycardia and WBC count were collected from the same day as the platelet count. We also documented available culture data, as well as treatments and patient outcomes.
To determine the etiology of thrombocytosis for each case, we reviewed discharge summaries and clinical notes, including daily progress notes, hematology consultations and infectious diseases consultations. We classified patients as having either primary or secondary thrombocytosis, then subcategorized patients by specific etiology as outlined below; our sources for the following definitions included prior case series of patients with thrombocytosis [2][3][4][5][6][7].

Definitions
Patients with primary thrombocytosis had a diagnosis of  gemcitabine, iron sulfate, ciprofloxacin, or piperacillin/tazobactam, as these drugs have previously been associated with thrombocytosis [8][9][10].
In terms of clinical outcomes, patients were classified as having either complete resolution (the underlying cause of thrombocytosis was resolved by the time of discharge), incomplete resolution (the underlying cause of thrombocytosis was still present at the time of discharge, or the patient was readmitted for the same condition within 1 month of discharge), death (the patient expired prior to discharge or within 1 month of discharge from a cause of death clearly related to the cause of thrombocytosis) or unknown (insufficient follow-up available). Patients with an infectious cause of thrombocytosis were classified as being managed with no treatment, with antibiotics alone, with surgery (if the patient underwent a procedure intended to treat the infection), or with a combination of antibiotics and surgery.

Timing of data collection
Data were collected between April and August 2006. Fol-low-up was extended for an additional 2.5 years for the 147 patients initially classified as having an indeterminate cause of thrombocytosis, which reduced that number to 59.

Statistical analysis
Data were entered into a Microsoft Excel database, and all analyses were conducted using SAS statistical software version 9.1 (SAS Institute, Cary, NC). Univariate comparisons of demographic and clinical characteristics of patients with primary and secondary thrombocytosis were conducted. Differences among categorical variables were tested using the chi-square test. Corresponding odds ratios (OR) and 95% confidence intervals (CI) were used to measure the likelihood of an infectious cause of thrombocytosis for various predictor variables. The percentage of patients with each possible etiology of thrombocytosis was calculated (primary, secondary non-infectious, and secondary infectious). Stepwise logistic regression was also performed with whether the etiology of thrombocytosis was infectious or non-infectious (including primary and secondary non-

Degree of thrombocytosis and time to normalization of platelet count
The platelet count for each subject was recorded as either high (500 -599 × 10 9 /L), very high (600 -800 × 10 9 /L) or extreme (> 800 × 10 9 /L). Most patients (53.8%) had a high platelet count, while 33.3% had a very high platelet count. Only 12.9% of subjects had extreme thrombocytosis. Although most patients' platelet count normalized within 6 months (63.6%), a substantial portion had unresolved thrombocytosis at the termination of the study (15.1%).

Etiology of thrombocytosis
Although the elevated platelet count was attributed to multiple causes in some patients, each etiology was considered separately for data analysis, thus bringing the total number of cases to 890 (Table 2). In most cases, thrombocytosis was secondary; primary causes of thrombocytosis were rare. The etiology of thrombocytosis was indeterminate in 7.4% of subjects (after re-review). Essential thrombocythemia (ET) was the most common cause of primary thrombocytosis. Among secondary, non-infectious etiologies, tissue damage was the most common, followed by malignancy and irondeficiency anemia. The most common infectious causes of thrombocytosis were soft-tissue, pulmonary and GI infections.
A variety of microorganisms were isolated from patients with infectious causes of thrombocytosis, although culture data were unavailable for some subjects. As shown in Table  3, the most frequently isolated microorganism was methicillin-resistant Staphylococcus aureus (MRSA), followed by Clostridium difficile. Other common microorganisms includ-ed Enterococcus sp., Pseudomonas sp. and Escherichia coli.

Characteristics of patients with secondary thrombocytosis due to infection
On univariate analysis (Table 4), subjects who were inpatients at the time of thrombocytosis were statistically more likely than outpatients to have an infectious etiology, as were patients with quadriplegia/paraplegia, an indwelling prosthesis, dementia or diabetes, as compared with patients without these characteristics. Vital signs and laboratory values associated with an infectious cause of thrombocytosis included fever, tachycardia, weight loss, hypoalbuminemia, neutrophilia, leukocytosis and anemia.
On multivariate analysis (using a stepwise logistic-regression model), the following characteristics were predictive of an infectious cause of thrombocytosis: fever, inpatient status, dementia, leukocytosis, diabetes and tachycardia (Table 5). Of note, weight loss, hypoalbuminemia, and elevated absolute neutrophil count were not included in the stepwise logistic-regression model due to missing values for more than 20% of patients; otherwise, all characteristics identified as significant on univariate analyses (Table 4) were entered into the model.

Etiology and degree of thrombocytosis
Patients with primary thrombocytosis had a higher mean platelet count (857.3 × 10 9 /L) than patients with secondary, non-infectious (654.1 × 10 9 /L) or infectious causes (644.2 × 10 9 /L). In fact, patients with platelet counts > 800 × 10 9 /L (extreme thrombocytosis) were more likely to have an MPD (OR, 4.75; 95% CI 2.45 -9.20; P < 0.0001). Apart from this finding, the degree of thrombocytosis did not correlate with the etiology of thrombocytosis or with a particular microorganism.

Etiology and time to normalization of platelet count
Patients with primary thrombocytosis were more likely to have a prolonged elevation in platelet count (time to normalization > 1 month) (OR, 8.86; 95% CI, 3.11 -25.21; P < 0.0001). Patients with secondary, non-infectious etiologies were also more likely to have a prolonged time to normalization (OR, 1.53; 95% CI, 1.13 -2.07; P = 0.0054). By contrast, patients with an infectious etiology were less likely to have prolonged thrombocytosis (OR, 0.35, 95% CI, 0.26 -0.48; P < 0.0001).

Clinical outcome
Most patients experienced either complete or partial resolution of the underlying cause of thrombocytosis (39.2% and 40.7%, respectively), but a minority of patients (7.9%) expired. In 12.2% of cases, the outcome was unknown, based on the medical record. Of patients with an infectious cause of thrombocytosis, 13.3% expired, compared with 4.4% of patients with non-infectious causes of thrombocytosis; in other words, patients with an infectious cause of thrombocytosis were more likely to die than patients with a non-infectious cause of thrombocytosis (OR, 3.34; 95% CI, 1.83 -6.09; P < 0.0001).
Because we had anecdotally observed that patients with an undrained focus of infection had prolonged clinical courses and persistent thrombocytosis, we investigated the relationship between treatment modality and clinical outcome. Of patients with an infectious cause of thrombocytosis, 62.7% of those treated with surgery had complete resolution at discharge, compared with 46.3% of patients who received antibiotics alone or no treatment; in other words, patients with surgical management of an infection were more likely to have complete resolution of the underlying infection (OR, 1.94; CI, 1.23 -3.09; P = 0.0043).

Discussion
In this study, we reviewed records of 801 patients with an elevated platelet count, with the goal of identifying clinical characteristics that point to an infectious cause of thrombocytosis. Our results corroborate prior findings [1][2][3][4][5][6][7] but also provide novel insights into the microbiology and clinical outcomes associated with infectious causes of thrombocytosis.
Secondary thrombocytosis was far more common than primary thrombocytosis, and the frequency of infectious and non-infectious causes was roughly equivalent. These findings are supported by prior investigations of adult patients with thrombocytosis [1][2][3][4][5][6][7].
Essential thrombocythemia was the most common MPD in this study, in keeping with at least 1 prior series [5]. Also consistent with prior studies, patients with a MPD had a higher mean platelet count, and patients with extreme thrombocytosis (> 800 × 10 9 /L) were more likely to have a MPD [1,3,7,11]. Although the mechanism of thrombocytosis in MPD is still under investigation, purported mechanisms involve genetic mutations that result in hypersensitivity to proteins that stimulate platelet production [12][13][14][15][16][17]. These genetic alterations may explain the more dramatic and prolonged elevation in platelet count associated with MPD.
Among secondary, non-infectious causes of thrombocytosis, tissue damage was the most common etiology, followed by malignancy and iron-deficiency anemia, again in keeping with prior series [2,4,6]. Patients with secondary, non-infectious etiologies were more likely to have prolonged thrombocytosis (> 1 month). This result likely reflects the chronic nature of many non-infectious etiologies (for example, malignancy, iron-deficiency anemia, chronic inflammatory disease).
Skin/soft-tissue and pulmonary infections were the most common infectious causes of thrombocytosis in this study, though GI and GU infections were also frequent. Similarly, a recent survey of thrombocytosis in an adult Turkish population showed that urinary tract and respiratory infections were the leading diagnoses, followed by "connective tissue" and GI infections [1]. A series by Griesshammer et al in 1999 showed pneumonia to be the most frequent infectious cause of thrombocytosis, while soft-tissue and GI infections were the next most common diagnoses [5]. A smaller study by Robbins and Barnard in 1983 showed pulmonary infections to be most common, followed by "surgical" infections [6].
To our knowledge, this is the first study of adult patients with thrombocytosis to provide information on etiologic microorganisms. MRSA and C. difficile were most commonly isolated in this study, which may be due to either a high prevalence of MRSA and C. difficile infections among the study population, or the relative ease with which MRSA and C. difficile can be identified with standard microbiological techniques. On the other hand, these organisms may have an increased propensity to stimulate reactive thrombocytosis. Interestingly, in a recent study of 162 patients with C. difficile, 22% were found to have an elevated platelet count, and toxin A was isolated more frequently in patients with both leukocytosis and thrombocytosis, suggesting that the pretest probability for C. difficile infection may be improved by thrombocytosis [18]. Convincingly, in vitro data suggest that the quantity of platelets is important for controlling S. aureus infection [19] and observational data in humans has shown that thrombocytopenia is associated with poorer outcome in S. aureus bacteremia [20].
Additional research would be required to investigate the tendency and/or mechanisms for MRSA and C. difficile to cause an elevated platelet count.
As in prior series [5][6][7], patients with infectious etiologies had a faster normalization of the platelet count, highlighting the acuity of infectious causes of thrombocytosis.
Notably, patients with an infectious etiology were also more likely to die in this study compared with patients with non-infectious causes of thrombocytosis. In fact, the mortality rate among patients with infectious etiologies was strikingly high (13.3%). Among patients with an infectious cause of thrombocytosis, those treated with surgery were more likely to have complete resolution at the time of discharge. Although the retrospective study design limits the interpretation of this finding, the data support our observations of patients with prolonged clinical courses and persistent thrombocytosis related to an undrained focus of infection.
The main limitations in this study are those associated with a retrospective investigation: the available data varied between patients, and we could not assess causative relationships. In addition, while the Michael E. DeBakey VA Medical Center serves a large population with a wide array of illnesses, the demographic homogeneity and primarily male population may have influenced our results.
Overall, our findings suggest that an infectious cause of thrombocytosis should be considered in patients at risk for infection (namely, inpatients or those with quadriplegia/ paraplegia, an indwelling prosthesis, dementia or diabetes) or in patients whose presenting complaints are suggestive of common infections (namely, skin, pulmonary, GI, GU). Clinical or laboratory evidence of severe illness (fever, tachycardia, weight loss, hypoalbuminemia, neutrophilia, leukocytosis or anemia) should also raise the index of suspicion for an infectious cause. Although most infections result in transient thrombocytosis, a persistently elevated platelet count should prompt a work up for an inadequately treated infection. Empiric treatment for soft-tissue and GI infections in patients with thrombocytosis may warrant coverage for MRSA and C. difficile, since these were the most commonly isolated microorganisms in this study. Appropriate surgical management of infection may be associated with a better outcome. Finally, since infectious causes of thrombocytosis were correlated with an increased risk of death, it seems prudent to recommend deciphering the cause of thrombocytosis and treating if possible.