Psoriasis vulgaris and familial cancer risk- a population-based study

Psoriasis is one of the most common skin disorders, and it is estimated that it affects 2-3% of the general Caucasian population[1]. Psoriasis is a chronic inflammatory skin disorder characterised by keratinocyte hyperproliferation and increased cutaneous blood flow induced via stimulation of tissue resident immune cells and a marked alteration of cytokine profiles[2]. Current evidence suggests that psoriasis is an immune-mediated disorder, and innovative therapies involved in the suppression of immune responses, such as T-cell-targeted agents and tumour necrosis factor (TNF) inhibitors, have improved the outcome of the disease[3, 4]. As judged by twin- and large population-based genetic studies, the disease seems to have a strong genetic component – disease concordance in monozygotic twins is at most 70%, and the sibling recurrence risk of PsV has been estimated to range between 4 and 11[5]. As of the publication of this writing, the molecular background of this disease remains unclear[6].

At present it is generally accepted that the disease is both multifactorial and genetically heterogeneous. GWAS studies identified several loci that are associated with psoriasis[7, 8]. Recently 15 new susceptibility loci (which included candidate genes whose products are involved in innate host defense) have been revealed, increasing to 36 the number associated with psoriasis in European individuals[9].

Follow-up studies of psoriasis patients point to an increased risk in the occurrence of malignancies at different sites of origin. Some reports suggest more common development of non-melanoma skin cancer among psoriasis patients[10‐13]. Results of other studies point to the increased frequency of other neoplasms, such as cancers of the larynx, lung, colon, kidney, pancreas and non-Hodgkin's lymphoma[12, 14‐16]. Unfortunately, data from literature is not entirely consistent. According to Alderson and Clark, cancers of the skin, stomach or lung are not associated with psoriasis[17]. In a recent study of Finnish psoriasis patients, the estimated relative risk was higher for Hodgkin’s and non-Hodgkin’s lymphomas, squamous skin cancer and laryngeal cancers; the risk for other malignancies (such as colon, lung or kidney) was unchanged when compared to the general population[18].

Exposure to UVA and UVB (used in the treatment of psoriasis) most likely increases the risk of malignant melanoma[19, 20]. According to Lee et al., melanoma and lymphoma are overrepresented among severe psoriasis patients[21]; however, the risk of malignant melanoma was reported to be diminished among psoriasis patients in other studies[15, 18].

Inconsistent data from literature can occur due to population stratification and/or complicated interpretation of evidence on the risk of cancer due to the small number of patients included in most series, precluding a full evaluation of rare neoplasms. Furthermore, the selection of patients in a clinical series might result in assessment of the effect of special types of therapies used in subsets of patients.

Herein we have investigated the risk of occurrence of malignancies of a different site of origin in a series of 517 psoriasis patients and their 1st degree relatives. We evaluated the tumour spectrum and age of diagnosis of cancers diagnosed in PsV families, as well as the observed and expected frequencies of malignancies.

Mutations/polymorphisms of many cancer susceptibility genes (such as BRCA1, CDKN2A, XPD, VDR, etc.) lead to tumour development. Some of these “cancer risk” gene alterations were also reported to be associated with psoriasis risk[22‐24]. To evaluate the possible link between psoriasis and cancer, we examined the distribution of 17 mutations/polymorphisms in 10 known cancer susceptibility genes among patients and 517 matched healthy controls.

No such study has been published to date.

In the first stage of the study, we compared frequencies of malignant tumour occurrence and the age of the patient at diagnosis in PsV families against those of the general Polish population. We evaluated malignancies affecting probands and their 1^st degree relatives.

For statistical analyses, the Chi-Square test, U-test and odds ratio were utilised. Additional analyses performed in PsV families included the comparison of observed frequencies (OF) with expected frequencies (EF) and the relative risk (RR) of occurrence of malignancies at different sites of origin. OF and EF were calculated through evaluation of the total number of family members and affected cases in different age groups (range of 5 years) in PsV families in comparison to age-specific incidence rates in different age groups (range of 5 years) per 100,000 people. These calculations were done by site with individuals registered in the general population of Poland[25]. Bonferroni correction was used for multiple testing.

In the second stage of the study, we compared the prevalence of common alterations in DNA reported in literature to cancer susceptibility alterations among PsV patients, healthy controls and the general Polish population (data obtained from published reports). We genotyped founder BRCA1 mutations (5382insC, C61G, 4153delA), common variants of XPD (D312M, K751Q), NOD2 (3020insC), CDKN2A (A148T), CHEK2 (CHIVS2, I157T), VDR (M1T), p53 (P72R), ATM (E1978X), MC1R (R160W, R151C, R163Q), MTHFR (A222V) and rs67 (rs6983267).

DNA samples were obtained from the peripheral blood of individuals. All SNPs were analysed by real-time PCR, using a LightCycler480 from Roche. The analyses were performed using the TaqMan(R) genotyping assay, which consists of sequence specific primers and oligonucleotide fluorescent labelled probes, enabling amplification of the examined fragments and further allele discrimination. Randomly selected probes were sequenced to confirm the results of real-time PCR.

Evaluation of the tumour spectrum revealed no statistically significant differences between proportions of malignancies observed in our PsV families when compared to the general population (Table 1). Analysis of the mean age of diagnosis (performed in our families for cancers affecting at least 5 cases) also revealed no significant differences between malignancies diagnosed among members of PsV families and the general population.

There was a tendency for an earlier appearance of leukaemia (~4 years), laryngeal cancer (~6 years) and melanoma (among males ~6 years) in PsV families (Table 1).

We observed 196 cancers among 3252 members of 517 psoriasis families. In comparison to the expected frequency (190 tumours), the difference was not significant (p=0.7931), neither for males (p=1.0) nor females (p=0.6519).

A statistical comparison of the observed and expected frequencies of cancers revealed a higher than expected occurrence of Hodgkin’s lymphoma among males in PsV families when compared to the general population (OR=1.8, 95%CI 1.6-2.1, p=0.002) (Table 2). There was a tendency, although non-significant, for melanoma overrepresentation among females from PsV families (OR=1.4, p=0.26). A similar tendency for both males and females was observed for leukaemia (OR=1.4, p=0.26) and laryngeal cancer (OR=1.3, p>0.19).

Molecular examination of 17 mutations/polymorphisms in 10 cancer susceptibility genes in psoriasis patients and controls showed no major differences in distribution of cancer susceptibility mutations among PsV families and healthy controls (Table 3).

For analysis of genetic markers, the statistical power a posteriori ranges between 0.12 in the most favourable case (marker rs6983267, variant GT) and 0.001 in the least favourable case (marker CHK2, variant IVS2+1G>A). Comparatively, for the most favourable case, the minimum sample size for a statistical power a priori of at least 0.7 should have been great than 1800 cases and an equivalent number of controls.

For analysis by cancer site, the statistical power a posteriori ranges between 0.17 in the most favourable case (female lung cancer) and 0.0015 in the least favourable case (several types of cancer, e.g. male liver cancer). Comparatively, for the most favourable case, the minimum sample size for a statistical power a priori of at least 0.7 should have been greater than 12400 first degree relatives.

Tumour spectrum and the age of the patient at diagnosis of malignancies of different sites of origin observed in our PsV families and in the general population were similar. However, there was a higher than expected frequency of the occurrence of Hodgkin’s lymphoma in males from PsV families. We also observed a tendency towards overrepresentation of leukaemia and laryngeal cancer among PsV families.

Our findings are consistent with the results of a follow up study of Swedish hospitalised psoriasis patients, which point to a significant excess of Hodgkin’s lymphoma and cancers of the upper aerodigestive tract[26]. An increased risk of Hodgkin’s lymphoma and laryngeal cancer has also been reported for Finnish hospitalised patients[18]. The association of leukaemia and psoriasis is not well-documented. There are a series of case reports of leukaemia developing in psoriasis patients treated with immunosuppressive drugs, such as cyclosporine, methotrexate or etanercept[27‐31]. Our results support the theory that psoriasis is not associated with an increased risk of NHL overall or of any NHL subtype[32], although there are some reports suggesting that such an association exists[16, 18]. We observed a non-significant tendency of melanoma overrepresentation among females from PsV families. This insignificant correlation does not clarify the inconsistent data from literature regarding the association of melanoma with psoriasis. We show that the risk of melanoma is not significantly increased for individuals from PsV families.

Due to the relatively small numbers of cases, both type 1 and 2 statistical errors cannot be excluded; the results thus need to be verified by examination of a larger series of patients. The current study lacks the statistical power to confirm if the absence of observed association is due to a true lack of association or rather due to an insufficient sample size. The current research should be considered only as an exploratory study attempting to reveal interesting tendencies worthy of in-depth analysis at a later stage.

Assessing the risk of cancer as a single outcome revealed no statistically significant association between psoriasis and cancer. It indicates that the common tumours (such as cancers of the breast, prostate, lung or colon) constituting the majority of neoplasms are not strongly linked with psoriasis. However, it does not exclude a moderate association for rare malignancies.

The results of molecular genotyping of mutations/polymorphisms present in cancer susceptibility genes revealed no differences in the prevalence of the examined alterations between cases and controls. It seems that none of the examined alterations in DNA are associated with psoriasis. This is the first such study evaluating any linkage between the BRCA1 gene (predisposing to breast and ovarian cancer), CHEK2 gene (cancers of the prostate, thyroid, breast, colon, stomach), MC1R gene (melanoma, non-melanoma skin cancers), ATM gene (cancers of the breast, pancreas, thyroid) and the XPD gene (melanoma, oesophageal squamous cell carcinoma, cancers of the lung and breast) with psoriasis. Our results are consistent with current data in literature, suggesting that NOD2 (colorectal and breast cancer), p53 (cancers of the cervix, bladder, prostate, breast) and MTHFR (oesophageal and cervical cancers) might not be susceptibility genes for psoriasis[33‐36]. Our findings support the thesis that the VDR gene (predisposing to melanoma and breast cancer) does not show a robust and reproducible association with the risk of psoriasis[37]. Since this theory conflicts with other findings[22, 23, 38], any association that may exist is likely to be weak and potentially restricted to specific populations. Down-expression of another cancer susceptibility gene, CDKN2A (predisposing to melanoma, cancers of the pancreas, breast and lung), led to speculation that the gene may have a role in the pathogenesis of psoriasis[24]. Our findings do not support this thesis.

According to many follow-up studies, there is an increased risk of the most common human malignancy, basal cell cancer (BCC), among psoriasis patients[10‐13]. In a recent study, genetic susceptibility to basal cell carcinoma (BCC) among Danish psoriatic patients was investigated and confirmed previous reports that XPD may predispose to BCC[39‐41]. Interestingly, we did not observe any excess of non-melanoma skin cancers in our families and found no association of XPD common variants with psoriasis. This lack of association suggests that the risk is restricted to individuals treated with phototherapy, which constituted only a small proportion of the individuals in our PsV families.

In conclusion, the results of our study indicate a moderate association between psoriasis and familial cancer risk. These results also suggest an increased risk of Hodgkin’s lymphoma for male members of PsV families. Further studies are needed to confirm the findings and to evaluate whether or not cancer surveillance protocols are justified in these families.