Hodgkin disease and testicular cancer are two forms of cancers that are commonly diagnosed in men of young reproductive age. While treatments result in high survival rates, questions about fertility and the possibility of fathering children post-treatment are of particular concern. Are there negative impacts due to the disease state itself and/or caused by the chemotherapy treatments that can affect the sperm? Can these have implications for the next generation? This study adds to previous work, including reports from our group, demonstrating an increased incidence of aneuploidy, DNA damage and altered chromatin quality in human sperm following combination chemotherapy treatments for HD and TC (7–10). Our findings here reveal distinct DNA methylation patterning in sperm prior to the onset of treatment that are able to distinguish HD, TC and CC subject groups from one another; interestingly, these profiles between the different groups remained post-treatment. As well, by comparing baseline and post-chemotherapy samples, while most initial defects appear to be resolved at later time points, altered sperm DNA methylation was detected in patients up to 18–24 months post-treatment, indicating that treatment induced epigenetic abnormalities, like genetic damage, persist long after patients become disease free.
Imprinted genes acquire sperm-specific DNA methylation patterns during male germ cell development and play crucial roles in normal development; disturbances in genomic imprinting can lead to a number of disorders (32). We analyzed the methylation of several imprinted genes represented on the 450K arrays and found normal profiles in the sperm of men diagnosed with HD or TC, prior to, and following, combination chemotherapy treatments. Our pre-treatment results are in keeping with those of a recent study examining the DNA methylation status of imprinted genes in cryopreserved sperm samples from normospermic before treatment seminoma patients and control subjects, where no alterations in imprinted gene methylation were found (33). The absence of imprinted gene DNA methylation abnormalities in testicular cancer patients prior to treatment in two different studies is reassuring. We are not aware of studies that have compared imprinted gene methylation in patients before and after treatment. However, we previously examined effects of human testicular cancer treatment regimens (BEP) on sperm DNA methylation in a rat model (17). Similar to our human studies, imprinted gene methylation was not affected by the BEP treatment. In a post-treatment genome-wide study of sperm DNA methylation in adult patients 10 years after pubertal exposure to chemotherapy for osteosarcoma, no imprinting gene methylation abnormalities were reported (6).
The study presented here is novel, in that we were able to examine the human sperm DNA methylome in samples acquired prior to the initiation of treatment and follow the same patients at several time points post-chemotherapy. Our results with the 450K arrays demonstrated that, prior to treatment, sperm DNA methylation signatures are present in men and are able to distinguish HD and, to a greater extent, TC patients from CC subjects. Interestingly, these group-specific DNA methylation patterns persisted in sperm following chemotherapy. Pre-treatment signatures were not evident using the lower resolution RLGS method. RLGS examines DNA methylation at a relatively small number of loci compared to the 450K arrays and can only reproducibly detect larger changes in DNA methylation, greater than 25% in magnitude. Similar to our findings, sperm aneuploidy has also been detected in pre-treatment samples from testicular cancer and Hodgkin disease patients (7, 34). Explanations for the presence of pre-treatment signatures include the impact of a cancer environment on germline stem cells, underlying genetic susceptibility, and an altered somatic environment. Recent studies examining the incidence of birth defects in the children of men with cancer, support the potential clinical relevance of the presence of pre-treatment genetic and epigenetic alterations in the spermatozoa of cancer patients (13). While studying sperm aneuploidy is labour intensive and requires specialized expertise, genome-wide DNA methylation assays are more widely accessible and may facilitate the screening of pre-treatment spermatozoa for prospective studies.
Our study suggests that pre-treatment DNA methylation differences between groups can persist post-treatment. In support of the finding in our prospective cohort, validation of a locus found to be altered in TC patients also showed greater variation in an external cohort, where samples were collected 2–3 years following chemotherapy. The most likely explanation would be persistence of abnormal germline stem cells or spermatogonial stem cells (SSCs) in the testis post-treatment. SSCs that had been quiescent may not have been eliminated by the chemotherapy. That SSCs with abnormal DNA methylation patterns can persist in the testis for many years is supported by a study of osteosarcoma patients treated as adolescents and examined as adults (6). Although pre-treatment samples were not available for the latter study, altered DNA methylation patterns were found in spermatozoa ten years after pubertal exposure to chemotherapy.
The pre-treatment aberrant epigenetic marks found in the sperm of HD and TC patients may identify a pre-existing signature that could be used to screen patients. Indeed, both diseases have etiologies associated with altered epigenetic patterning. The distinctive cells used for the diagnosis of HD, the Hodgkin/Reed-Sternberg cells, have been shown to have lost or decreased many of the B cell-specific gene expression profiles (35, 36). By altering DNA methylation and/or histone acetylation patterning, studies in cell lines have been able to alter the expression patterns of B cells to resemble more closely those characteristics of HD, and vice versa (37–39). Testicular cancer, hypospadias, cryptorchidism and poor semen quality, are included in the clinical entity, testicular dysgenesis syndrome, a disorder postulated to be due to environmental effects that disrupt embryonic and fetal germ cell development (40, 41). Indeed, the precursor of testicular germ cell tumors, the carcinoma in situ (CIS) cell, has been theorized to originate from fetal gonocytes, through morphological and gene expression profiling (42, 43). Epigenetic marks of the CIS have also been demonstrated to be similar to those of fetal germ cells (44, 45), allowing for increased proliferative activity. Interestingly, biological processes relating to the immune system/immune response and sexual/male reproduction were observed in the differentially methylated regions from the baseline analysis of the HD and TC cohorts, respectively, alluding to the origins of these diseases. While the altered epigenetic signatures from our patient cohorts were discovered in sperm, it remains to be determined whether more accessible biological samples, such as blood or saliva, would also carry perturbed epigenetic patterning.
From cancer studies, methylation of loci identified in both HD and TC patient sperm are known to influence gene expression. The largest differentially methylated region found in HD patients was within the SPON2 locus, showing increased methylation in HD versus CC individuals. Altered methylation within the promoter of this extracellular matrix protein has been observed both in prostate cancer and in cancer-associated myofibroblast cell lines from gastric and oesophageal adenocarcinomas compared to normal tissues (46, 47). Decreased methylation within the promoter led to a corresponding increase in SPON2 expression within the cell lines, which was also detected in patients with the poorer prognosis and survival of gastric/colorectal carcinoma patients (48, 49). Specific to testicular cancer, decreased methylation in sperm was observed within the promoter of GDF2 in our patient cohort, a result also seen in a subset of TC patients 2–3 years post-treatment in the validation cohort. This locus encodes the GDF2 protein, also known as bone morphogenic protein 9 (BMP9), a member of the TGFβ superfamily, which has roles in cancer progression (50). Specifically, GDF2 activation of SMAD1/5 signaling has been found to enhance cell death in ovarian and breast cancer cell lines; treatment of these cells with 5-azacytidine, a potent demethylating agent, demonstrated decreased methylation of the GDF2 locus, increased signaling and higher sensitivity to cell death (51). This same study also observed higher GDF2 promoter methylation in tumors of ovarian cancer patients compared to normal individuals. Similarly, loss of GDF2/BMP9, in a mouse model of breast cancer, was associated with increased tumor growth and metastasis, indicating a protective role of this gene (52). From the cancer studies, the increased methylation of SPON2 in the sperm of HD patients would result in decreased expression while the decreased methylation of GDF2 in the sperm of the TC patients would result in increased expression. We hypothesize that such altered expression patterns, perhaps involving other genes identified in the HD and TC cohorts, may provide a growth or differentiation advantage for germ cells.
Immediately following chemotherapy treatments, many patients in the HD and TC cohorts showed an elevated number of sites with altered sperm DNA methylation compared to the CC groups. Indeed, a significant increase in DNA damage, as measured by tail extent moment (Comet assay) was detected in a similar group of both HD and TC patients 6 months after treatments (9). We also observed that many of the initial defects observed were resolved and the number of altered sites diminished with time, similar to the treatment-induced damage to sperm chromatin structure being differentially repaired over time in cancer survivors. However, one patient demonstrated increased alterations in the sperm methylome in the following two years; similar significant sperm DNA damage and low DNA compaction has been observed up to 24 months post-treatment (10). It is also intriguing that developmental and nervous system pathways were enriched in examining our post-treatment effects for GO biological process terms. Taken together, alterations in sperm DNA methylation, altered sperm chromatin structure and DNA damage can be observed in HD and TC patients prior to and following chemotherapy regimens. This damage to the genetic/epigenetic cargo of spermatozoa may help to explain the observed increased risk of congenital abnormalities in children fathered by men with a history of cancer, whether the offspring were conceived prior to or years following treatments (11, 12).
One limitation of our study is the small number of patients available in each cohort, as well as the reduced numbers of samples available following treatments. While these are a common form of cancers in men of reproductive, the incidence is still low, with 2.8 and 6.8 per 100,000 in the United States for HD and TC, respectively (53, 54). As well, the impact of treatments on spermatogenesis is evident with decreased sperm counts immediately following chemotherapy, which only return to pretreatment levels after 24 months (55). Despite the low patient numbers in our different cohorts, we were able to detect differential methylation signatures in sperm. As mentioned previously, a validation in other cohorts of HD and TC patients of these signatures would be beneficial. Another limitation is that we were only able to examine a subset of the ~ 32.8 million possible CpG methylation sites found in the human genome (56). The majority of the results presented here are from the Illumina 450K array, examining approximately 450,000 CpG sites, with these mainly covering CpG islands and over gene regions. Our early study examining sperm DNA methylation using the 450K array, showed that the majority of CpGs interrogated (86% of probes) were hypo- or hypermethylated (< 20% or > 80%, respectively) with methylation levels conserved between individuals; only a small proportion of probes (0.3%) were variable between individuals and demonstrated intermediate (20–80%) methylation (57). We reported similar finding (i.e. a minority of variable CpG sites) in a more recent study using the 450K array and examining the effects of low dose folic acid supplementation on sperm DNA methylation(58). Our use of the human sperm-specific MMC-Seq that examines 3.18 million sites, did demonstrate a greater number of alterations in patients who underwent chemotherapy treatments; the vast majority of the altered DNA methylation was found in dynamic CpGs, sites of intermediate methylation and not well covered on the different Illumina arrays. Indeed, the use of our sperm-specific panel for the analysis of human DNA methylation differences due to a genetic polymorphism, environmental exposures, as well as the process of ageing, has demonstrated that differential methylation is mainly found in these novel dynamic CpG sites targeted (31, 59). Future studies using our human sperm targeted panel, or with whole genome bisulfite sequencing, may be useful to examine in greater detail the effect of cancer as well as their treatments on the sperm methylome.