To The Editor: I read with particular interest the manuscript on “Venous Thromboembolism Prevention in Cancer Outpatients” (J Natl Comp Canc Netw 2013;11:1431-1438). The author stated that clinicians interacting with outpatients with cancer should recommend outpatient thromboprophylaxis to select patients, including those at high risk based on the risk score, those with pancreas cancer initiating chemotherapy, and those with myeloma initiating IMid- (immunomodulatory drug) based combination regimens, in the absence of contraindications.” He also noted that post hoc subgroup analysis of the PROTECHT study population with high risk score (≥3) revealed that the rate of venous thromboembolism (VTE) was 11.1% in the placebo arm and 4.5% in the nadroparin arm, and thus the number needed to treat (NNT) to prevent 1 VTE event was 15. Another reference cited was the SAVE-ONCO trial, in which VTE events in the high-risk group were 5.4% in the placebo arm versus 1.4% in the semuloparin arm (NNT, 25). The author also acknowledged that in patients with pancreatic cancer, low-molecular-weight heparin (LMWH) prophylaxis may require higher-than-typical prophylactic dosing to achieve a substantial VTE reduction benefit.¹

Although I agree with ambulatory thromboprophylaxis (A-TP) in patients with myeloma who are receiving IMid-based combination regimens due to the high incidence of VTE events without A-TP, I have concerns about the A-TP recommendation in patients with pancreatic cancer and in those with solid cancers with high risk scores (≥3). I believe these recommendations are premature for several reasons.

First, there have been many randomized, controlled A-TP trials in patients with solid cancer receiving outpatient chemotherapy. None of them have shown statistically significant reduction in fatal VTE events or overall survival.^2,3 Second, in the PROSPECT-CONKO-004 trial in patients with pancreatic cancer, the enoxaparin arm received 1 mg/kg/d for 3 months followed by 40 mg/d for an additional 3 months.⁴ For a 60-kg patient, this translates to an enoxaparin requirement of 9000 mg over 6 months. Because this study has not been published yet, we do not know the NNT to prevent 1 VTE event. If we also look at the FRAGEM study carefully, the dalteparin arm received 200 IU/kg/d for weeks 1 to 4 and 150 IU/kg/d for weeks 5 through 12. For a 60-kg patient, this translates to a dalteparin requirement of 840,000 IU over 12 weeks. The FRAGEM study is a small phase IIb study, and therefore it is difficult to reliably calculate the NNT to prevent 1 VTE event. Even for 1 patient, LMWH expense is large.⁵ Further, many questions remain unanswered, including the optimal dosing for A-TP in this population; whether to use enoxaparin, 40 mg/d or dalteparin, 5000 U/d or enoxaparin, 1 mg/kg/d or dalteparin, 200 U/kg/d; and the optimal duration (3 or 6 months; the duration of chemotherapy; or beyond the duration of chemotherapy). Other questions include what is the correct NNT to prevent 1 VTE event in patients with pancreatic cancer and what is the cost/benefit ratio. Finally, if patients with pancreatic cancer have additional risk factors (eg, body mass index >35 kg/m² or hemoglobin level <10 g/dL) and a risk score of 3 or greater, do they have equivalent or higher risk of VTE than patients with other solid tumors with risk score 3 or greater?

Nadroparin is not available in the United States and therefore we cannot calculate its expense to prevent 1 VTE event. As the NNT to prevent 1 VTE event in the high-risk population was 15 in the PROTECHT trial, its cost would probably be pretty high. Theoretically speaking, if semuloparin had been approved in the United States based on the SAVE-ONCO trial, its cost to prevent 1 VTE event in patients at high risk would still have been very high because the NNT to prevent 1 VTE event was 25 in this trial.¹

In the recently reported FRAGMATIC trial, the largest randomized A-TP trial in patients with lung cancer (1100 patients enrolled in each arm), the addition of 24 weeks of treatment with dalteparin (5000 U/d) to chemotherapy did not lead to improved overall survival (hazard ratio, 0.97; 95% CI 0.89-1.06), although a modest reduction in VTE rates was seen (7.1% vs 4.1% in the dalteparin group).⁶

We must also consider the potential problems with subset analysis such as multiple comparison problems, overinterpretation of observed differences, and apparent interaction effects. Subset analyses provide indirect evidence and are subject to substantial bias.⁷ An example is the routine recommendation of thromboprophylaxis in inpatients with cancer by major oncology organizations based on extrapolated data from inpatients in general medicine. A recent study by Carrier et al⁸ showed a lack of evidence to support thromboprophylaxis in inpatients with cancer.

Finally, 17.6% of the gross domestic product of the United States is spent on health care.⁹ A-TP will undoubtedly increase health care costs unless a truly significant cost-effective A-TP regimen can be implemented. Perhaps a better question is whether a strategy of educating patients with cancer about VTE risk and promptly recognizing and treating VTE once it develops is more cost-effective than using A-TP in patients with solid cancers at high risk. Only carefully designed, randomized, controlled studies will be able to answer this question. Those studies are ongoing and the results are eagerly awaited.