Incidence of Venous Thromboembolism in cancer patients treated with Cisplatin based chemotherapy — a cohort study

There is a substantial risk of venous thromboembolic events (VTE) in patients with cancer. The incidence of VTE in the general population is approximately 117 per 100,000, whereas the incidence in patients with cancer is approximately one in 200 [1, 2]. A large cohort study showed that the baseline risk of thrombosis in cancer is approximately 4.1 folds compared to the healthy population while it increases to 6.5 fold after the addition of chemotherapy [3]. This leads to reduction in survival of patients with malignancy and increased expense of treatment and care [4‐6]. Although individual factors including the tumor type, stage, recency of cancer diagnosis, presence and number of co-morbidities define risk of VTE in cancer, active therapy with certain chemotherapeutic agents has been contemplated as an additional risk factor [7]. There is anecdotal evidence of increased incidence of VTE in patients receiving Cisplatin based regimes [8, 9]. Studies in urothelial, germ cell and non-small cell lung cancer (NSCLC) patients have established a link between Cisplatin and increased incidence of VTE [10‐12]. Another recent large retrospective study showed the prevalence of VTE to be 16.6% in all patients treated with Cisplatin [13] which is significantly higher than 7.3% for VTE in patients treated with any chemotherapy [14]. Prospective randomized controlled trials in patients with advanced esophagogastric junction (EGJ) or gastric adenocarcinoma showed that this thrombogenic effect is not class specific for all the platinum containing chemotherapeutic agents, but only Cisplatin based chemotherapy is associated with a statistically significant greater number of VTE events than Oxaliplatin [15, 16]. Recently, a meta-analysis from case series and some pre-clinical trials provided evidence of high risk of VTE in patients receiving Cisplatin based regimens [17].

Various hypotheses exist pertaining to the mechanisms associated with Cisplatin associated thrombosis including direct damage to the vascular endothelium, increased procoagulant activity, reduced anticoagulation synthesis, platelet activation and aggregation and vascular inflammation [18‐24].

Despite increasing evidence pertaining to the thrombogenic potential of Cisplatin, no original analytical study has thus far investigated the magnitude of increased risk of VTE imparted by Cisplatin based chemotherapies when compared to other chemotherapies. Hence, it is extremely important to quantify this risk in order to advocate prophylaxis with antiplatelets or anticoagulants and hence reduce the economic and health care associated burden of VTE caused by Cisplatin based chemotherapy. Therefore we aimed to quantify the risk and identify high risk groups so that treatment is targeted and benefits maximized. Prophylactic anticoagulation for all cancer patients is expensive and has a narrow therapeutic index and hence not practical. We hypothesized that there is an increased risk of VTE in cancer patients receiving Cisplatin based chemotherapy regimens compared to other chemotherapies.

This is a retrospective cohort study including all adult cancer patients presenting to the oncology ward or day care centre of the Aga Khan University Hospital for chemotherapy. Cancer patients receiving Cisplatin based chemotherapy regimens were considered exposed while those cancer patients receiving chemotherapy regimens which do not contain Cisplatin comprised the unexposed group. Adult (>16 years) participants were considered eligible if they received treatment between November 1, 2010 and October 31, 2012. They should have initiated and completed all treatment at Aga Khan University Hospital while maintaining a followup of atleast 1 month post completion of treatment or until death, whichever came first. Excluded were those on antiplatelets agents or anticoagulants, with a previous history of DVT or PE (self-reported), known pro-thrombotic disorders or hyper-coagulable states, patients who were on erythropoiesis-stimulating agent (ESA) during chemotherapy or up to 6 weeks prior to the first cycle of chemotherapy and those female patients who were on oral contraceptive pills.

All cancer patients treated in the day care centre or the inpatient oncology wards of Aga Khan University Hospital between November 1, 2010 and October 31, 2012 were identified through the discharge coding system. 100 consecutive eligible patients were included in each group starting from November 1, 2010 (Fig. 1). Medical record files were reviewed for demographic and clinical data which was recorded on a pre-designed questionnaire. Confounding factors such as the presence of indwelling central line, obesity and family history of DVT were recorded if present. All identified patients were followed up during the course of chemotherapy and at least 4 weeks post chemotherapy. Radiological investigations for the study population were also reviewed to make sure that no cases of VTE are missed. In case of discrepancy between the chart data and radiological investigations, the patient was telephonically evaluated for the history of VTE event or the use of anticoagulant agents during the specified time period. Outcome was any VTE event during chemotherapy or within 4 weeks after the completion of chemotherapy. ECOG classification was modified for analysis as class 0, 1 and 2 were combined as “good”, while class 3 and 4 was “poor”. Class 5 was excluded. Khorana score was categorized as low risk (0 points), intermediate risk (1–2 points) and high risk (3 or more points).

The mean age of Cisplatin group was 56(11.9) years while in the non-Cisplatin groups it was 55(9) years. There was a slight preponderance of males in both groups (Table 1). Most of the cancers were metastatic in both groups and very few were early stage malignancies (Tables 2 and 3). Less than 10% of the whole cohort had ECOG poor status. 129(64.5%) participants in Cisplatin group had intermediate Khorana risk score as compared to 125(62.5%) in the Non-Cisplatin group. 29% of the patients had an indwelling central venous catheter (CVC) at some point during the study in the Cisplatin group vs. 27% in the non-Cisplatin group. Details of the chemotherapy regimens used in both groups have been highlighted in Tables 4 and 5 whereas important co-morbids in the two cohorts have been summarized in Table 6.

When the dose of Cisplatin was considered, we found that 31 VTE events occurred when the mean (SD) cumulative dose of Cisplatin was 471(133) mg/m² while in the group without events the mean cumulative dose was 322(124) mg/m² (Table 7).

Among the VTE events most were DVTs i.e.18 in Cisplatin group while 7 in non-Cisplatin group (Appendix 1). Among those who suffered a VTE event, 4 died in Cisplatin group while 1 in non-Cisplatin group (Appendix 2). On univariate analysis other than Cisplatin based chemotherapy, poor ECOG status, presence of CVC and Khorana risk score were statistically significant (Appendix 3). These were used for final model building.

The crude relative risk of VTE in the Cisplatin group was 2.8 (95% CI: 1.4–4.22) times higher than in the Non-Cisplatin group (Table 8). When adjusted by gender, ECOG status, GCSF, presence of CVC and Khorana risk score the adjusted relative risk was 3.32(95% CI:1.6–6.8) (Table 9).

This study reports a high risk (RR: 3.3, 95% CI: 1.6–6.8) of VTE in cancer patients receiving Cisplatin based chemotherapy. Cisplatin is a platinum based drug which is used for the treatment of cancer. The mechanism of causing VTE is not clear but endothelial injury, hypomagnesemia and increased levels of Von Willebrand’s factor are involved [8, 19‐21, 23]. Patient factors including female sex, age, type of cancer, Khorana risk and indwelling CVC further elevate the risk of VTE [13]. In our study Khorana risk and presence of CVC were found to be significantly associated with this risk. Another interesting finding, which has not been explored previously, is the risk of VTE with higher cumulative doses of Cisplatin (Table 3). In our study, cumulative Cisplatin dose of more than 450 mg was associated with VTE events.

Our risk estimates are higher than a previously reported meta-analysis where the risk was 1.67 (95% CI: 1.25–2.23). The reason for this may be Seng et al, in their meta-analysis pooled results from phase II to III clinical trials. It is possible that patients in their study were a highly selective group. Another explanation could that we had highly specialised investigations available and the chances of diagnosing a VTE are higher in the present era than between 1990 and 2010 which is the study time for Seng et al’s study [17].

The reason for choosing one month follow-up was twofold. One, previous literature [13] had also taken one month follow-up for development of DVT in patients receiving Cisplatin based chemotherapy. This study reported that 88% of patients developed TEE within 100 days of initiation of treatment with Cisplatin based regimens. Hence a period of 30 days post completion of chemotherapy seems a reasonable follow-up period for capturing most, if not all, events occurring as a result of the drug intervention. Secondly, we believe it is safe to assume that since re-dosing of Cisplatin as a chemotherapeutic agent is required within 3–4 weeks of the last dose; most therapeutic as well as adverse effects of the drug are neutralised within one month of last dosing. Moreover, if a VTE occurs one month after completion of chemotherapy, it becomes further complex to attribute it solely to chemotherapy effects as underlying disease and other risk factors (e.g. CVC, immobility etc.) are likely to play a much more important role.

Our major strength is being the first original research reporting relative risk of VTE in Cisplatin based regimens compared to non-Cisplatin. Aga Khan University hospital is one of the two large centres for the treatment of cancer in Pakistan. Hence, we were able to obtain a reasonable sample with reliable data on patients from the time of starting Cisplatin therapy up till one month post chemotherapy. We were able to maintain close follow up with all our patients without attrition. There was no missing information because we had access to medical records and patients were coming for clinical follow ups regularly. We chose a retrospective study design for feasibility of sample size achievement and save resources.

The data was collected retrospectively but since all the parameters were measured and documented in hospital notes we assume there is minimal bias due to recall. Although we may have missed some important confounders, one advantage of using hospital data in this study was that the medical details were very accurate and radiological investigations were easy to follow hence minimizing misclassification of exposure and outcome.

Site of cancer could be an important confounder hence it was incorporated as one of the predictors of Khorana score and it is reflected within the scoring. Accordingly, we have adjusted our final model for Khorana score to eliminate any confounding effect that may have arisen due to the site of disease.

Polychemotherapy may potentially attribute higher rates of VTE than single agent chemotherapy but this fact is unlikely to affect our results as almost all patients in both the groups received combination chemotherapy (Tables 4 and 5). Patients who received other platinum based therapies (Carboplatin, Oxaliplatin), were included in the non-Cisplatin group further highlighting the possibility that the increased thrombogenic effect may be inherent to Cisplatin alone rather than to the entire group of platinum drugs.

This however, is also a potential bias and limitation of our study as a difference in VTE incidence between the different platinum drugs was not explored further. Our study had not been designed to evaluate for any differences in VTE incidence between the platinum compounds but further dissection and analysis could indeed prove very interesting and is an area which needs to be explored in future studies.

Another potential limitation of our study is the absence of a universal screening test for assessment of VTE status prior to initiation of chemotherapy in patients. Although this would have been the ideal method to ensure absence of VTE (target end-point) in newly recruited patients, the retrospective nature of our study meant that we had no control over investigations performed both prior to and during the course of the study.

In the light of these finding we think there is an urgent need to prospectively estimate the risk of VTE in patients receiving Cisplatin based regimens. It is immensely important to discover the pathophysiology and identify genetic and molecular mechanisms of occurrence of VTE in these patients. Till then we strongly recommend prophylaxis against VTE in high risk groups to avoid serious morbidity or mortality. We emphasize that it is extremely important to develop potent and effective chemotherapeutic agents with fewer side effects.

In our experience this is one of the first reports of risk estimates of VTE with Cisplatin based chemotherapeutic regimens. Focus should be laid on discovery of newer and safer chemotherapeutic agents. Till then, prophylaxis against VTE for patients receiving Cisplatin based chemotherapy is highly recommended. It is important to identify high risk groups vulnerable to this side effect.