Design Flaws In Statins And Pancreatic Cancer Research (Vol 145, Pg 1448, 2020)

We read with great interest the recently published study on the association between statins and pancreatic cancer risk in patients with chronic pancreatitis (CP).1 Preventing the progression from CP to pancreatic cancer is of great significance. Statins are widely used as cholesterol-lowering drugs nowadays. Therefore, it is of great importance to identify whether statin is related to lower risk of pancreatic cancer in CP patients. However, research design needs to be improved. First, the present study just use 1 year of follow-up after the diagnosis of CP to reduce the risk of misdiagnosing pancreatic cancer as CP. It is difficult to diagnose pancreatic cancer at its early stage, and early stage pancreatic cancer is often misdiagnosed as CP. From our point of view, patients with suspected pancreatic cancer should be excluded from the current study, and a period of 1 year is insufficient for excluding patients with misdiagnosed pancreatic cancer.2 In almost all CP research, particularly involving CP and pancreatic cancer, patients diagnosed with pancreatic cancer within 2 years of follow-up were excluded and some studies even used another 5 years to mitigate misdiagnoses further.3-5 Second, the main treatments for CP include endoscopic therapy and surgery. However, the present study did not mention about the treatment of CP patients. According to a previous Danish study,6 29.7% CP patients had undergone pancreatic surgery. It is reported by a multicenter retrospective study that CP patients who underwent surgery had significantly lower incidence of pancreatic cancer (hazard ratio estimated by Cox regression = 0.11, 95% confidence interval: 0.0014–0.80, p = 0.03).5 Therefore, whether CP patients had underwent surgery is a nonnegligible confounding factor which affects the association between statins and pancreatic cancer in CP patients. Treatment method of CP population should be adjusted in the multivariable models. Third, Gagne Comorbidity Index was adjusted as a composite measure of overall comorbidity burden of each patient. However, Gagne Comorbidity Index is an index which only concerns about mortality in patients. The prevalence of both cardiovascular and cerebrovascular disease is significant higher in statin ever users compared to statin never users (Table 1, p < 0.001). Patients with cardiovascular or cerebrovascular disease may have a higher proportion of smokers. Smoking doubles the risk of pancreatic cancer.3 Therefore, Gagne Comorbidity Index could not represent the bias caused by the two comorbidities in this study. Impacts of differences in comorbidities for pancreatic cancer between statin ever users and statin never users may have been weakened by only adjusting Gagne Comorbidity Index. Hence, these two comorbidities should be adjusted in the multivariable analysis. Fourth, statins is a drug with frequent drug–drug interactions (DDIs).7, 8 DDIs between statins and drugs prescribed for comorbidities of CP may interfere efficacy of statins which can cause bias in the result. However, the author did not analyze the complex drug usage in these patients. The neglected drugs for comorbidities like cardiovascular disease (e.g., warfarin) have impact on statins’ metabolism and efficacy.9 Even in the drugs listed in the present study, statins have obvious DDIs with nonsteroidal antiinflammatory drugs, proton pump inhibitors, antidiabetic drugs and different kinds of statins according to previous reports.7, 8 In addition, the prevalence of potential statin–drug interactions in older people (>65 years) was reported to be up to 33.0%.10 Older people occupied 24.9% of statin ever users. As a result, drug usage for the CP patients in the present study is complicated as various DDIs exist and the statins’ efficacy may be affected. Therefore, it would be better if the author could adjust the complex drug usage in multivariable analysis to reduce bias. In conclusion, identifying the association between statins and pancreatic cancer in CP patients is valuable. The present study should be improved and this may help to develop a more accurate instruction for statins usage in CP patients. Yours sincerely