Association of hydrochlorothiazide use and risk of malignant melanoma
Menée au Danemark à partir de données portant sur 19 273 patients atteints d'un mélanome malin et sur 192 730 témoins, cette étude évalue l'association entre une utilisation d'hydrochlorothiazide, une molécule utilisée comme médicament diurétique de la classe des thiazidiques, et le risque de développer la maladie
We have recently shown that hydrochlorothiazide use increases the risk of lip and nonmelanoma skin cancer, notably squamous cell carcinoma.1,2 It would have substantial implications if the carcinogenic effect of hydrochlorothiazide also extended to malignant melanoma. Methods : Similarly to our recent studies of hydrochlorothiazide,1,2 we identified histologically verified melanoma cases (January 2004 to December 2015), each matched 1:10 (risk-set sampling; age, sex, and date) to cancer-free population controls. We required cases and controls to be between ages 18 and 90 years, without previous history of cancer (except nonmelanoma skin cancer), organ transplantation, HIV infection, or azathioprine use, and to have resided continuously in Denmark for 10 years. Using conditional logistic regression, we calculated odds ratios (ORs), with 95% CIs, for melanoma associated with cumulative hydrochlorothiazide use compared with never-use, adjusting for potential confounders (Table 1 and 2). We performed stratified analyses by localization, stage, histologic subtype, and subgroups of age, sex, and history of nonmelanoma skin cancer. To evaluate potential confounding by indication, we performed analyses for other antihypertensive drugs, including bendroflumethiazide, angiotensin-converting enzyme inhibitors, angiontensin-II receptor antagonists, and calcium-channel blockers. This study was approved by Statistics Denmark and the Danish Data Protection Agency. Results : We identified 22 010 cases of melanoma. After exclusions, the final study population comprised 19 273 cases and 192 730 population controls. Cases had slightly lower comorbidity, higher educational level, and higher prevalence of previous nonmelanoma skin cancer than controls. Remaining characteristics were similar between cases and controls. Overall, 413 cases (2.1%) and 3406 controls (1.8%) were classified as high-users (≥50 000 mg) of hydrochlorothiazide, yielding an adjusted OR of 1.22 (95% CI, 1.09-1.36) for melanoma. No clear dose-response pattern emerged between hydrochlorothiazide use and melanoma risk (Table 1). Analyses by melanoma localization, stage, age, sex, and history of nonmelanoma skin cancer yielded results comparable to the main analysis (data not shown). When stratifying by histological subtype (Table 2), higher ORs occurred for nodular melanoma (n = 1695 cases [8.8%]; OR, 2.05; 95% CI, 1.54-2.72; P for trend = .01) and lentigo melanoma (n = 500 cases [2.6%]; OR, 1.61; 95% CI, 1.03-2.50; P for trend = .16) than for superficial spreading melanoma (n = 13 781 cases [72%]; OR, 1.11; 95% CI, 0.97-1.27; P for trend = .73). In secondary analyses, we observed associations close to the null for overall melanoma risk with long-term use of bendroflumethiazide (OR, 1.10; 95% CI, 1.02-1.19; P for trend = 0.47), angiotensin-converting enzyme inhibitors (OR, 1.07; 95% CI, 0.99-1.16; P for trend = .53), angiotensin-II receptor antagonists (OR, 1.18; 95% CI, 1.07-1.29; P for trend = .07), and calcium-channel blockers (OR, 1.06; 95% CI, 0.97-1.14; P for trend = .94). These associations remained neutral in subanalyses stratified by melanoma subtype (data not shown). Discussion : The main strength of our study is the use of high-quality nationwide registry data.3 The main limitations are a lack of information on risk factors such as sun exposure, skin pigmentation, and family history of melanoma. However, these characteristics are unlikely to be substantially associated with hydrochlorothiazide use, and thus unlikely to confound out estimates. Thiazide use and melanoma risk has been investigated in a few previous studies; however, only 2 studies,4,5 both from northern Denmark, have specifically examined hydrochlorothiazide. The first study reported an OR of 1.32 (95% CI, 1.03-1.70) for melanoma risk overall associated with 10 000 mg increments of hydrochlorothiazide.4 The corresponding OR for hydrochlorothiazide in combination with amiloride was 1.43 (95% CI, 1.09-1.88).4 The other study found no association between hydrochlorothiazide use combined with amiloride and melanoma risk (OR, 1.02; 95% CI, 0.78-1.33).5 Neither of these studies included dose-response or histology-specific analyses. The findings for melanoma subtype are somewhat surprising, as lentigo and superficial spreading melanoma are known to be associated with high sun exposure, whereas the etiology of nodular melanomas is less elucidated.6 It is worrying that hydrochlorothiazide use appears to be associated with an increased risk of melanoma, and the particular associations observed for lentigo melanoma and nodular melanoma warrant further research.