Thrombocytosis as a prognostic factor for survival in patients with metastatic renal cell carcinoma

Abstract

BACKGROUND.

It has been suggested that thrombocytosis, as defined by a platelet count >400,000/μL, is a negative predictor for survival among patients with metastatic renal cell carcinoma. However, this has not been a uniform finding.

METHODS.

To address this issue, retrospective analysis of 700 previously untreated patients entering on institution review board-approved phase 1, 2, or 3 clinical trials in the United States and Europe was conducted between 1982 and 2002.

RESULTS.

Thrombocytosis was present at study entry in 25% of patients. Median baseline platelet count was 304,000/μL (range, 86–1,420,000/μL). Eighty-seven percent of patients died with a median survival of 13.0 months. Median follow-up for patients not known to have died was 2.4 years. On univariate analysis, patients with elevated platelet counts had significantly shorter survival than patients with normal platelet counts; median survivals of 8.4 and 14.6 months, respectively, P < .001. However, platelet count was associated with several clinical and biochemical factors, including gender, age, performance status, time from diagnosis to study entry, prior radiotherapy or nephrectomy, presence of liver metastasis, number of metastatic sites, amount of hemoglobin, white blood cell count, amount of lactate dehydrogenase, and amount of serum alkaline phosphatase. Several of these factors have previously been reported as prognostic indicators for survival, and, therefore, multivariable analyses were conducted to determine whether thrombocytosis is an independent predictor of survival. After adjusting for multiple factors, thrombocytosis continued to impact negatively on survival, P < .001.

CONCLUSIONS.

Thrombocytosis was found to be an independent prognostic factor for survival in patients with metastatic renal cell carcinoma. Cancer 2006. © 2006 American Cancer Society.

Renal cell carcinoma (RCC) accounts for 36,000 new cancer diagnoses yearly in the United States, constituting 2.5% of all malignancies and 2% of cancer mortality.1 One-third of patients present with metastatic disease at initial diagnosis. Fifty percent of surgically resected patients will ultimately relapse. The median survival for metastatic RCC remains poor at 12–15 months, with 5-year overall survival 0–20%.

The standard therapy for metastatic RCC has been immunotherapy with response rates of only 10–15%.2, 3 Chemotherapy has minimal, if any, role in these patients.4 A growing understanding of the underlying biology of RCC has led to development of vascular endothelial growth factor (VEGF) inhibitors, such as SU11248 (Sunitinib) and BAY 43-9006 (Sorafenib). The promising data with VEGF inhibition in metastatic RCC have established new opportunities for improving outcomes in this historically resistant malignancy.5, 6 However, several questions remain unanswered concerning their optimal use.

Understanding prognostic factors can provide insights into disease biology, identify patient populations likely to benefit from a particular treatment, aid in clinical trial interpretation, and stratify patients for various therapies. Several poor prognostic factors have been identified in advanced RCC patients, such as low performance status, high serum lactate dehydrogenase (LDH), low hemoglobin (Hb), high serum calcium, time from diagnosis to treatment of less than 1 year, absence of prior nephrectomy, history of prior radiotherapy, and number of metastatic sites.7–10 Several models have been developed that combine these factors to define prognostic patient subgroups. One recent example frequently used in this setting is the Memorial Sloan Kettering Cancer Center (MSKCC) model developed by Motzer et al.10, 11 The MSKCC model defines 3 risk groups based on the number of poor prognostic factors present (Karnofsky performance status <80%, time from diagnosis to treatment <12 months, corrected serum calcium >10 mg/dL, serum Hb below the lower limit of normal, and LDH >1.5 times the upper limit of normal). Patients considered to have a favorable profile are those with no poor prognostic factors present; intermediate patients have 1–2 factors present; and patients with an unfavorable profile have >2 factors present.

Secondary or reactive thrombocytosis occurs in several clinical conditions, including various cancers12 such as metastatic renal cell carcinoma.12, 13 Thrombocytosis, defined by a platelet count >400,000/μL, was first documented as a negative predictor for survival in metastatic RCC by Symbas et al.14 However, this has not been a consistent finding.15 To address this issue, we retrospectively reviewed the charts and laboratory results of 700 previously untreated metastatic RCC patients enrolled on phase 1, 2, or 3 clinical trials conducted in the United States and Europe.

MATERIALS AND METHODS

Records of 700 consecutive patients with metastatic RCC enrolled on institutional review board-approved clinical trials between 1982 and 2002 at the Cleveland Clinic Foundation in Cleveland, Ohio, Memorial Sloan Kettering Hospital in New York City, New York, and Centre Leon Berard in Lyon, France, were reviewed. The trials were primarily phase 1, 2, or 3 studies of immunotherapy with or without chemotherapy or other investigational agents. All patients signed an informed consent form before enrollment in the trial. Eligibility criteria for the trials were fairly uniform and generally included the following: histologic documentation of RCC, clinical or biopsy evidence of metastatic disease, bidimensionally measurable disease, Eastern Cooperative Oncology Group (ECOG) performance status of ≤1, adequate bone marrow, renal, and hepatic functions, absent CNS metastases, no prior history of other cancers (except basal cell or carcinoma-in-situ of the cervix that was curatively treated), no significant cardiac disease, and no recent surgery.

Factors evaluated in patients included the following: medical center where treatment was performed, age, gender, diagnosis to on-study, performance status, history of prior nephrectomy, prior radiation therapy, presence of lung, liver, osseous, mediastinal, retroperitoneal, or other metastases, number of metastatic sites, Hb, white blood cell count (WBC), platelet count, calcium, LDH, alkaline phosphatase levels, and MSKCC risk group. Platelet count was obtained before the start of the treatment protocol as part of the normal workup for study entry. It was this pretreatment count that was used to classify patients by the presence or absence of thrombocytosis. Thrombocytosis was defined as >400,000/μL. Patients who initially had platelet counts <400,000/μL and later developed thrombocytosis were categorized as having a normal count and were included in the analysis.

Statistical Analysis

Thrombocytosis was summarized using frequency counts. For convenience, biochemical parameters and factors such as age and time from diagnosis to study entry were categorized using recursive partitioning methods or cut points previously reported in the literature. The chi-square test, Fisher exact test, and the Cochran–Armitage trend test were used to compare patient groups with the frequency of thrombocytosis. Survival was measured from the start of treatment to death or last follow-up. Survival distributions were estimated using the Kaplan–Meier method. The relation of survival to patient and disease factors was analyzed using the log-rank test and trend test. Multiple factors were simultaneously assessed using the Cox proportional hazards model with stepwise variable selection.

RESULTS

Patient Characteristics

Seven hundred previously untreated patients were enrolled on trials from 1982 to 2002. Of the 700 patients whose medical charts were reviewed, thrombocytosis was present at study entry in 25% of patients. Patients received a variety of single agents and combination therapies; the majority received interferon alpha (IFN-α) alone (21%) or regimens based on IFN-α (68%). Seventy-one percent of patients were males, and median age was 57.9 years. Seventy-seven percent of patients had a prior nephrectomy. Median baseline platelet count was 304,000/μL (86–1420). Median follow-up for patients not known to have died was 2.4 years and 87% of patients were known to have died. The median survival of the 700 patients analyzed was 13 months.

Patient characteristics, the prevalence of thrombocytosis at entry on study, and univariate survival analyses are summarized in Table 1. Thrombocytosis was associated with several factors, including younger age (<60 years), a short interval from diagnosis to treatment, ECOG performance status 1 or 2, history of prior nephrectomy or radiotherapy, liver metastasis, low Hb, elevated WBC or alkaline phosphatase, and increasing MSKCC risk group. The prevalence of thrombocytosis was also associated with participating center.

Table 1. Patient and Disease Characteristics, Thrombocytosis, and Univariate Survival

Factor	No. (%)	% with platelets >400 K/μL	P*	No. deaths (%)	Median survival (mo)	Hazard ratio (95% CI)†	P‡
CCF indicates Cleveland Clinic Foundation; MSKCC, Memorial Sloan-Kettering Cancer Center; mo, months. * Fisher exact test unless otherwise noted. † Hazard ratio of the second level of a factor relative to the first, unless otherwise noted. ‡ Log-rank test unless otherwise noted. § Compared with CCF. ‖ chi-square test. ¶ Cochran-Armitage trend test. # Hazard ratio treating the factor as ordinal. ** Trend test. †† LLN (ULN), lower (upper) limit of laboratory's reference range. ‡‡ Corrected calcium could not be calculated for 38% of patients because albumin was not collected; consequently, total serum calcium ≤ or >11 mg/dL was used instead (corrected and total calcium are highly correlated, Pearson r = 0.90; 11 mg/dL was the cutoff used for assessing total serum calcium in the original analysis2).
Thrombocytosis
No	522 (75)	—	—	439 (84)	16.4
Yes	192 (25)	—	—	171 (96)	8.4	2.06 (1.72, 2.46)	<.001
Center
CCF	150 (21)	20		112 (75)	17.3
MSKCC	288 (41)	32		274 (95)	12.6	1.25 (1.00, 1.56)§
Centre Leon Berard	262 (37)	21	.004‖	224 (85)	12.2	1.26 (1.01, 1.58)§	.09
Age
<60	397 (57)	30		345 (87)	13.1
≥60	303 (43)	20	.003	265 (87)	13.0	1.05 (0.89, 1.23)	.59
Gender
Male	500 (71)	24		430 (86)	13.6
Female	200 (29)	30	.08	180 (90)	12.2	1.13 (0.95, 1.34)	.18
Diagnosis to on-study
≥1 y	253 (36)	18		208 (82)	18.9
<1 y	447 (64)	30	<.001	402 (90)	11.5	1.66 (1.40, 1.97)	<.001
ECOG PS
0	267 (38)	18		216 (81)	18.8
1	380 (54)	30		347 (91)	11.7
2	53 (8)	30	<.001¶	47 (89)	7.5	1.48 (1.30, 1.69)#	<.001**
Prior nephrectomy
No	160 (23)	36		154 (96)	9.5
Yes	540 (77)	22	.001	456 (84)	15.0	0.54 (0.42, 0.66)	<.001
Prior radiation
No	565 (81)	27		486 (86)	13.4
Yes	135 (19)	17	.01	124 (92)	12.1	1.17 (0.96, 1.43)	.12
Lung metastasis
No	196 (28)	22		174 (89)	14.8
Yes	504 (72)	27	.21	436 (87)	12.7	1.01 (0.85, 1.20)	.92
Liver metastasis
No	574 (82)	24		487 (85)	14.8
Yes	186 (18)	34	.02	123 (97)	7.8	1.74 (1.43, 2.12)	<.001
Bone metastasis
No	508 (73)	27		437 (86)	14.0
Yes	192 (27)	20	.06	173 (98)	11.7	1.28 (1.07, 1.52)	.01
Mediastinal metastasis
No	486 (69)	26		416 (84)	13.6
Yes	214 (31)	24	.71	194 (91)	12.6	1.16 (0.98, 1.38)	.09
Retroperitoneal metastasis
No	510 (73)	24		443 (87)	14.4
Yes	190 (27)	29	.20	167 (88)	10.4	1.25 (1.05, 1.50)	.01
Other metastasis
No	495 (71)	26		433 (87)	13.6
Yes	205 (29)	23	.34	177 (86)	12.2	1.01 (0.85, 1.20)	.92
No. of metastases
0, 1	228 (33)	25		195 (86)	15.1
>1	472 (67)	26	.93	415 (88)	12.2	1.21 (1.02, 1.43)	.03
Hemoglobin
≥LLN††	357 (51)	12		296 (83)	19.0
<LLN	343 (49)	40	<.001	314 (92)	9.7	1.92 (1.63, 2.25)	<.001
White blood cells (K/μL)
≤11.0	632 (90)	22		548 (87)	14.0
>11.0	68 (10)	60	<.001	62 (91)	6.3	1.64 (1.26, 2.14)	<.001
Calcium (mg/dL)
≤11	667 (95)	25		578 (87)	13.6
>11	33 (5)	27	.84	32 (97)	7.2	2.22 (1.55, 3.17)	<.001
Lactate dehyrogenase
≤1.5 X ULN††	626 (89)	24		544 (87)	14.4
>1.5 X ULN	74 (11)	34	.09	66 (89)	4.8	2.48 (1.91, 3.21)	<.001
Alkaline phosphatase (g/dL)
≤100	348 (50)	16		295 (85)	16.6
>100	352 (50)	35	<.001	315 (89)	11.6	1.48 (1.26, 1.73)	<.001
MSKCC risk group‡‡
Favorable	138 (20)	7		108 (78)	24.8
Intermediate	488 (70)	28		433 (89)	12.8
Unfavorable	74 (11)	46	<.001¶	69 (93)	4.8	2.12 (1.81, 2.48)#	<.001**

In univariate analysis, patients with thrombocytosis had a significantly shorter survival than patients with normal platelet counts. Median survival was 8.4 months and 14.6 months, respectively (P < .001). Additional factors that were also associated with poor survival were short time from diagnosis, increasing ECOG performance status, prior nephrectomy, the presence of liver or bone metastasis, retroperitoneal nodal involvement, more than 1 metastatic site, low Hb, elevated WBC, serum calcium, LDH, or alkaline phosphatase, and increasing MSKCC risk group.

Because of the large number of factors that were associated with thrombocytosis and/or survival, multivariable analyses were conducted to determine whether thrombocytosis was an independent predictor. The results of this analysis are summarized in Table 2. All factors in Table 1 were initially considered. By using stepwise variable selection, thrombocytosis, MSKCC risk group, age, number of metastatic sites, and alkaline phosphatase were found to be independent predictors of survival. Survival curves according to baseline platelet count are given in Figure 1.

Figure 1. Survival by platelet count.

Table 2. Multivariate Survival

Factor	Estimate + SE	Hazard ratio (95% CI)	P*
SE, standard error; CI, confidence interval; MSKCC, Memorial Sloan-Kettering Cancer Center; Alk phos, alkaline phosphatase; Met, metastatic. * Wald test.
MSKCC risk group	0.67 ± 0.08	1.96 (1.66, 2.31)	<.001
Platelets >400 K/μL	0.50 ± 0.10	1.65 (1.36, 1.99)	<.001
Age >60 y	0.27 ± 0.09	1.31 (1.11, 1.54)	.002
Alk phos >100 U/L	0.21 ± 0.08	1.23 (1.05, 1.45)	.01
>1 Met site	0.21 ± 0.09	1.23 (1.04, 1.46)	.02

DISCUSSION

Although advances in the treatment of metastatic RCC have been made in recent years, the overall outcome of this disease remains dismal. Despite encouraging results from new treatment agents, their optimal incorporation into clinical practice remains to be defined. Whether these agents should be used as monotherapy or combined with cytokines or other agents remains speculative. The role of prognostic factors, specifically platelet count, may help better define these questions.

We sought to analyze all metastatic RCC patients who were treated on various clinical trials at 3 academic centers. The objective of this study was to determine whether an elevated platelet count is a negative predictor for survival in metastatic RCC patients. Platelet count is a simple, inexpensive test that can be obtained before treatment and could help individualize therapy based on risk factor assessment. Our results showed that 25% of patients had thrombocytosis at study entry and that elevated platelet count remained an independent prognostic factor on multivariate analyses.

In the retrospective study by Symbas et al.,14 147 of 259 (57%) metastatic RCC patients were found to have at least 1 platelet count of >400,000/μL after nephrectomy. Mean survival for these patients was 92 months, compared with 151 months for those with normal platelet count (P = .005). Controlling for known prognostic indicators such as pathologic stage, nuclear grade and cell type, the difference in survival between the two groups remained significant (P = .015). This represented a 64% increase in survival time for patients who maintained a normal platelet count after being diagnosed with metastatic RCC. Conclusions from this study were that thrombocytosis may be a manifestation of aggressive tumors, with worse survival when compared with patients with normal platelet counts. The higher percentage of patients with thrombocytosis in this study is most likely attributed to the assessment of platelet count in the postoperative setting.

Of 204 patients with localized RCC (6.6% metastatic) at Emory University Hospital,16 26 (13%) were found to have platelet counts >400,000/μL.16 Of these, 23 (11%) were detected preoperatively and 3 (2%) postoperatively. A statistically significant (P < .001) difference was detected in overall and cause-specific mortality between patients without thrombocytosis (15.2% and 7.3%, respectively) and those with thrombocytosis (50% and 42%, respectively). Of the 23 patients with thrombocytosis preoperatively, the platelet count decreased in 14 after radical nephrectomy. In the patients with perioperative thrombocytosis who underwent nephrectomy for early stage RCC, the cancer-specific death rate from RCC was 5 times greater than the rate in patients with a persistently normal platelet count after surgery. The conclusion from this study was that platelet count is a powerful independent prognostic factor among patients with localized or advanced RCC, even when tumor stage, grade, and histologic subtype are controlled.

In another retrospective review of 196 patients treated by radical nephrectomy for RCC, thrombocytosis was evaluated before surgery.17 All stages were included; however, the majority (64%) was TNM Stage 1. Sixteen (8.2%) patients were found to have thrombocytosis (>400,000/μL) preoperatively. Platelet count normalized after nephrectomy in all 16 patients; there was no prognostic difference in the time of normalization. However, these patients had a worse prognosis than those with normal preoperative platelet counts. The mean number of months until death due to cancer was 13.9 months for the patients with thrombocytosis and 36.1 months for those with normal platelet counts (P = .0028). When adjusted for stage or tumor size, the worse prognosis correlated to only Stages 1 and 2 (P = .0141) or to tumor size ≤4 cm (P = .0021) or 4–7 cm (P = .0142).

Gogus et al.18 performed a retrospective review on 151 patients who underwent radical nephrectomy for localized RCC at Ankara University Hospital in Turkey. The platelet count was assessed only preoperatively. Twenty-one (14%) patients had a platelet count >400,000/μL. No statistically significant correlation was found between grade and histologic subtypes and thrombocytosis. However, advanced T stage (P = .045) and lymph node involvement (P = .028) correlated with thrombocytosis. Ten of 21 (47.6%) patients with thrombocytosis and 21 of the 130 (16.2%) patients with normal platelet counts died of disease progression (P = .002). Patients with thrombocytosis had a cause-specific mean survival of 45.2 months compared with 76.6 months for those with normal platelet counts (P = .0002). When controlled for stage, lymph node status, grade, and cell type, the difference in cause-specific survival between the two groups remained statistically significant (P = .023). The authors concluded that preoperative thrombocytosis is an independent prognostic factor for predicting cancer-specific survival in RCC patients.

In a French study of more than 700 patients treated on multicenter trials of cytokines,15 thrombocytosis was found to be a significant predictor for survival on univariate analysis. However, on multivariate analysis, 10 other factors were found to be predictive of survival outcome as follows: biologic signs of inflammation, performance status, hemoglobin, number of metastatic sites, bone metastases, neutrophil count, alkaline phosphatase level, liver metastases, mediastinal metastases, and time interval from renal tumor to metastases. Platelet count was not significant on multivariate analysis, possibly because of the masking effect of biologic signs of inflammation, which included erythrocyte sedimentation rate and C-reactive protein.

The exact mechanism producing thrombocytosis in association with RCC is unclear. Reactive or secondary thrombocytosis associated with malignancies has been established since the early 1870s,12 with an incidence of 10–57%.19–21 Possible mechanisms include an overproduction of cytokines/growth factors stimulating megakaryocytes and their precursors. In vitro experimental studies demonstrate that interleukins (IL), such as IL-6, IL-1, and leukemia inhibitory factor possess abilities to cause thrombocytosis.22–24 The correlation with thrombocytosis, metastatic RCC, and serum IL-6 levels was initially described by Hollen et al.25 Serum IL-6 is increased in most patients with reactive thrombocytosis,25 and elevation of this cytokine has been detected in a significant number of patients with RCC.26 More than half of patients with metastatic RCC have increased levels of circulating IL-6, which also correlates with increased C-reactive protein levels.26–28 In a study by Blay et al.,26 serum samples for IL-6, blood cell counts, and various other parameters were collected before chemotherapy and/or immunotherapy in 119 patients with metastatic RCC between 1988 and 1992. Of 119 patients, 90 (76%) had detectable serum IL-6 (median, 14 pg/mL; range, 1–1850 pg/mL) and 14 (12%) had levels greater than 80 pg/mL. Platelet counts were significantly higher (>450,000/μL) among patients with detectable serum IL-6. IL-6 levels greater than 80 pg/mL was associated with thrombocytosis (P = .0001).

In a study by Walther et al.,29 IL-6 was detected in 19 of 21 (90%) renal cancer cell lines obtained from 20 patients with metastatic RCC and also detected in the serum of 33 of 59 (56%) patients with metastatic RCC. Elevation of the cytokine was associated with paraneoplastic manifestations, such as anemia, thrombocytosis, decreased albumin, and elevated alkaline phosphatase. There was no correlation between pretreatment serum IL-6 level and survival or response to IL-2.

Thrombopoietin (TPO) is the key hormone in megakaryocyte differentiation and proliferation and is predominantly produced in the liver.30 Bone marrow endothelial cells, kidney, and spleen are also capable of TPO production.31, 32 TPO is produced and released into the circulation at a constant rate by the liver. Normal physiology of platelet production involves the clearance of TPO by high affinity TPO receptors on platelets and formation of a steady TPO concentration, thereby providing a basal stimulation of bone marrow megakaryocytes and normal rate of platelet production. However, in secondary thrombocytosis that can occur with malignancies, there can be up-regulation of TPO production by the liver, causing enhanced thrombopoiesis. Plasma TPO levels have also been shown to correlate with IL-6. In a phase 1 trial where patients were treated with recombinant IL-6, pretreatment TPO levels were below detection limit, and, subsequently, the levels significantly increased with highest concentration on posttherapy Day 5.33

Questions remain unanswered as to why thrombocytosis is associated with decreased survival in RCC patients. Several theories have been proposed on how platelets themselves have a significant role in tumor growth. Perhaps IL-6 and other cytokines, along with inflammation as a general phenomenon have an effect on the immune system and, thereby, result in worse survival.

Platelets have capabilities to enhance sequestration, adherence, and penetration of malignant cells through the endothelial wall. They may also prevent the immune system from clearing tumor cells from the circulatory system.34 Thrombospondin, a platelet-secreted protein, is increased in patients with cancer, specifically in patients with metastasis,35 and may promote the adherence of tumor cells to the endothelial barrier, thus enhancing their escape from immune surveillance.36

Tumor growth is dependent on formation of new blood vessels from preexisting capillaries (i.e., angiogenesis).37 Tumor angiogenesis is dependent not only on endothelial cells and cancer cells but also on platelet–endothelium interactions. Platelets adhere to the tumor-related endothelium and release high concentrations of VEGF,38 which is a potent stimulator of angiogenesis.39 A strong positive correlation has been detected between serum VEGF levels and platelet counts in RCC patients.40 Furthermore, a positive correlation between serum VEGF levels and disease progression was discovered in patients with advanced colorectal cancer.41

Platelet granules contain a variety of factors such as VEGF, basic fibroblast growth factor (bFGF), platelet-derived growth factor (PDGF), transforming growth factor beta (TGF-β), IL-6, thrombin, and fibrinogen. These modulators are secreted immediately after platelet activation, and many have been implicated in various steps of tumor progression and in the development of metastasis.42 As one of the most significant proangiogenic cytokines, bFGF contributes to migration, proliferation, and differentiation of endothelial cells, and regulation of the expression of proangiogenic molecules.43 PDGF induces angiogenesis by means of stimulation of VEGF expression in tumor endothelial cells and by recruiting pericytes to new blood vessels.44 TGF-β has an active role in platelet aggregation and regulation of megakaryocyte activity. This cytokine also regulates the activity of the VEGF system and enhances endothelial cell survival.45, 46

In conclusion, our study confirms the prevalence of thrombocytosis associated with metastatic RCC. We have also demonstrated that a platelet count >400,000/μL portrays a worse survival, independent of other clinical or biochemical factors. With further studies, this single independent prognostic factor may provide a simple approach to improved risk stratification of patients in future clinical trial protocols. The International Kidney Cancer Working Group has an on-going project to examine multiple prognostic factors and to devise a common approach to stratify RCC patients by risk.