br To further evaluate the potential pleiotropic effect of
To further evaluate the potential pleiotropic effect of the genetic instruments used in this study, we performed mendelian randomization–Egger regression Journal of Thoracic Oncology Vol. 14 No. 1
analysis. However, we observed no evidence of bias (p ¼ 0.66) (see Supplementary Figs. 3 and 4), and the direction of bias that we estimated was negative, sug-gesting that the effect of the putative bias present across the selected genetic variants resulted in an underesti-mation of the causal effect on lung cancer risk (see Supplementary Fig. 4).
Chromosome Y has long been recognized for its necessity in sex determination and male sperm cell development.15–17 However, genomic characterization of chromosome Y still lags far behind that of the rest of the GW3965 because it is considered to be a wasteland. Recently, as a male-specific genomic event, mLOY was found to frequently occur in aging men1 and shown to be associated with smoking status and increased hypodip-loidy.1,18–20 Population-based studies have also identi-fied associations between mLOY and hematological disorders and nonhematological malignancies2,4–6; however, these results are inconsistent and may be confounded by reverse causality.4–6 In this study, we have provided what to our knowledge is the first evi-dence for a causal relationship between mLOY and lung cancer based on the concept of mendelian randomiza-tion. Sensitivity analyses with alternative casual infer-ence methods demonstrated a consistent association, suggesting the robustness of our results. We propose that mLOY is a protective factor against both the tumorigenesis and progression of lung cancer, whereas cigarette smoking may weaken this effect. Together with the results of previous studies, our findings support the idea that chromosome Y carries many vital functions in biological processes beyond sex determination and sperm production.
In this study, one of the most noteworthy findings was a U-shaped curve between the copy number level of chromosome Y and lung cancer risk, whereas genetically defined mLOY was linearly associated with a decreased lung cancer risk. Additionally, we have also provided the first evidence that increased genetically predicted mLOY can lead to better prognosis of lung cancer. This result suggests that genetically defined mLOY is different from mLOY caused by environmental factors. Although genetically defined mLOY may promote benign aneu-ploidy and exert a protective factor against lung cancer, both gain and loss of chromosome Y caused by acquired damaging environmental factors may contribute to spontaneous tumorigenesis. As a well-known risk factor for lung cancer,21 tobacco carcinogen is reported to in-crease copy number alterations and large-scale chro-mosomal instability in lung cancer,22–24 and smoking can also induce mLOY in the blood in a dose-dependent and
January 2019 mLOY and Lung Cancer Risk in Chinese 43
transient manner.8 Thus, we believe that the effect of cigarette smoking may bias the estimation of mLOY.
Given the aforementioned results, we thoroughly tested whether cigarette smoking confounded the effect of mLOY on lung carcinogenesis. Interestingly, the protective effect of mLOY against lung cancer was observed in lifelong nonsmokers but not in smokers (pHeterogeneity ¼ 3.83 10–2) (see Table 1). These find-ings suggest that the aberrant loss of chromosome Y occurs more frequently in smokers than in nonsmokers and exerts an effect on lung cancer that is the opposite of that of genetically defined mLOY. Therefore, cigarette smoking may account for most of the environmental damaging effects of mLOY during the process of lung carcinogenesis. Additionally, on the basis of the assumption of mendelian randomization, all of the mLOY-associated genetic variants used in this study were not associated with smoking behavior, and hence, the bias caused by smoking status could be avoided. On the basis of the aforementioned hypothesis, it is reasonable to observe a relatively weaker protective effect of mLOY against lung cancer than that observed by Zhou et al.6 because of the relatively low smoking rate in the Asian population than in Europeans.25 Although genetically predicted mLOY was found to be a protective factor against lung cancer, the effect can be easily attenuated by external risk factors, which may account for the higher incidence and mortality of most non–sex-specific cancers in males. However, the phenotypic consequence of mLOY in the development of lung cancer is rather elusive and further functional studies in both cellular and animal systems are warranted to elucidate the exact mechanism. As cancers are heterogeneous in their origin and may be caused by different forms of exposure, we believe that the inconsistent associations found between mLOY and cancer risk in previous studies can be attributed to the underestimation of caner het-erogeneity and the effect of environmental factors.