Second primary cancers after anogenital, skin, oral, esophageal and rectal cancers: Etiological links?
Version of Record online: 19 APR 2001
Copyright © 2001 Wiley-Liss, Inc.
International Journal of Cancer
Volume 93, Issue 2, pages 294–298, 15 July 2001
How to Cite
Hemminki, K., Jiang, Y. and Dong, C. (2001), Second primary cancers after anogenital, skin, oral, esophageal and rectal cancers: Etiological links?. Int. J. Cancer, 93: 294–298. doi: 10.1002/ijc.1319
- Issue online: 9 JUN 2001
- Version of Record online: 19 APR 2001
- Manuscript Accepted: 9 FEB 2001
- Manuscript Revised: 16 JAN 2001
- Manuscript Received: 24 NOV 2000
- Cancer Society
- King Gustaf V′s Jubilee Fund
- oral cancer;
- human papillomavirus;
- cervical cancer;
- rectal cancer;
- anal cancer;
- genital cancer
The Swedish Family-Cancer Database was used to analyze second cancers after oral, esophageal, rectal, cervical, genital and skin (squamous cell carcinoma) cancers. A strong and consistent association of second cancers was observed at all these sites, in men and women. As a novel finding, an association of rectal cancer with the human papillomavirus (HVP)-related cancers was shown. New evidence on an excess of skin cancer with the HPV-related cancers was also provided. As an epidemiological study, the associations were strong and often supported by a number of comparisons. These could not be explained by bias or long-term treatment related effects. However, whether the findings on rectal and skin cancer are due to HPV or other infections, transient or inherited depressed immune function or other constitutional factors remains to be established. © 2001 Wiley-Liss, Inc.
A large proportion of patients diagnosed with a first primary cancer develop a second primary cancer, particularly in cancers of good survival. The interest in second cancers may be of two kinds: first, second cancers may indicate the effects of treatment for the first cancer, and second, they may be caused by the same environmental or genetic factors that caused the first cancer.1 In some cases the intensive medical surveillance after the first cancer may lead to overdiagnostics. However, in the Swedish Cancer Registry practically all reported cancers are histologically or cytologically verified, ensuring correctness of diagnosis. Human papillomavirus (HPV) is the main cause of cervical cancer and an important cause of other anogenital cancers including those of the anus, vulva and penis and, less convincingly, those of the head, neck and esophagus.2–8 Recent studies have provided molecular and epidemiological evidence for the role of HPV in head and neck cancers.9–12 The possible role of HPV in squamous cell carcinoma of the skin (here “skin cancer”) has remained unclear.2, 6, 13 In follow-up studies after cervical cancer, skin cancer has been found in moderate excess, and, conversely, in follow-up studies after skin cancer, genital cancers have been in excess.14–16 However, there may be other mechanisms for these associations, such as other infective or life-style factors, depressed immune function or genetic factors. There is no evidence linking HPV to rectal cancer; however, its proximity to the anus and the known potential of HPV to cause adenocarcinomas, as well as squamous cell carcinomas, may provide a mechanistic hypothesis.
We analyzed systematically the occurrence of second cancers in men and women who had been diagnosed with any of the following primary invasive cancers: oral (here including cancers of the lip, mouth, tongue, pharynx and larynx), esophageal, rectal, anal, cervical, other female and male genital and skin cancers. The rates for second cancers at these sites were adjusted for age, period, residential area and socio-economic status. In order to differentiate the effects of treatment, the rates of second cancer were assessed after various periods of follow-up.
SUBJECTS AND METHODS
The Swedish Family-Cancer Database includes practically all persons born in Sweden after 1934 with their biological parents, totalling over 9 million individuals.17, 18 Invasive cancers were retrieved from the nation-wide Swedish Cancer Registry from the years 1958 to 1996. A 4-digit diagnostic code according to the 7th revision of the International Classification of Diseases (ICD-7) was used. The following ICD-7 codes were pooled: “oral” cancer codes 161 (larynx) and 140 to 148 (lip, mouth, tongue, pharynx), except for code 142 (salivary glands). Rectal cancer, ICD-7 code 154, was separated for the anus (squamous cell carcinoma, 154.1) and mucosal rectum (154.0). Basal cell carcinoma of the skin is not registered in the Cancer Registry; thus squamous cell carcinoma is the only recorded skin cancer, termed “skin cancer” in the present paper. The Cancer Registry assigns notifications of cancer as the first, second, etc. primary cancer using identical criteria.
The numbers of recorded male and female first cancers were 10,780 and 3366 for oral, 14,449 and 10,064 for rectal, 334 and 744 for anal, 17,234 for cervical, 2528 for other female genital, 1127 for male genital and 11,084 and 6196 for skin cancers, respectively. Standardized incidence ratios (SIRs) were calculated by dividing the observed numbers (O) of second events by the expected (E) ones, calculated as person-years at risk based on age (5-year age groups)-, period (5-year periods)-, area of residence (two categories)-, socio-economic status (four categories)- and sex-specific incidence rates of all persons.19 Confidence intervals (95%CIs) were calculated assuming a Poisson distribution.19
After the first male oral cancer, an excess of oral (SIR 10.16), esophageal (9.31), rectal (2.18), genital (3.70) and skin cancers (5.54) was observed (Table I). For oral and esophageal cancer, the SIRs were very high, 38.29 and 33.39, respectively, within the year of diagnosis. At other sites, the follow-up time had no effect. Esophageal cancer has a poor survival, and only oral and esophageal cancers were increased as second cancers following esophageal cancer. After rectal cancer, oral and skin cancers were increased. After anal cancer, rectal and, particularly, genital cancers (SIR 60.24 but only two cases) were increased. After genital cancer, rectal and genital cancers were increased. After skin cancer, cancers at all sites were increased. The SIR for second skin cancer was as high as 15.92. It should be pointed out that at some sites significant increases were observed in two ways, e.g., first oral cancer led to an increase in rectal cancer and vice versa, and first rectal cancer led to an increase in oral cancer. Similar associations were noted between oral-skin and rectal-skin pairs of sites.
|First cancer||Second cancer||Follow-up interval, years|
|O||E||SIR||95% CI||O||E||SIR||95% CI||O||E||SIR||95% CI||O||E||SIR||95% CI|
Similar analyses were carried out for second cancers among women (Table II). The pattern of second cancers after oral cancer resembled the results from males. The SIR for second oral cancer of 108.46 within the first year of follow-up was the second highest found in this study. The highest SIR, 123.88, was for oral cancer after esophageal cancer. Only genital cancers were increased after rectal cancer; oral and skin cancers were increased after anal cancer. Cervical cancer was followed by an excess of oral, rectal, anal and other genital cancers. The highest risks after cervical cancer were observed in early follow-up. Other genital cancers were followed by an excess of oral, anal and genital cancers. Skin cancer resulted in an excess of oral, genital and skin cancers. Some of the associations between sites were confirmed both ways: oral-esophagus, esophagus-oral; oral-genital and genital-oral; oral-skin and skin-oral.
|First cancer||Second cancer||Follow-up interval, years|
|O||E||SIR||95% CI||O||E||SIR||95% CI||O||E||SIR||95% CI||O||E||SIR||95% CI|
We analyzed a group of cancer sites systematically for second cancers in order to find etiological clues for the aggregation of cancers between these sites. The cancer sites were selected based on a priori knowledge on HPV-related sites, or a related hypothesis on rectum and skin. With systematic analysis we understand the comparison of the results between men and women, and between two sites in two ways, i.e., we follow genital cancer after oral cancers, and vice versa, oral cancers after genital cancers. Because of the different age-incidence distributions of specific cancers, the results in the two-way analysis may not agree. However, if they do, they argue strongly for a true association. Moreover, it should be pointed out that in this kind of comparison the findings should be viewed beyond the formal statistical significance. For example, in women the association anal-genital (first-second cancer) showed an SIR of 4.73 but it was not significant (n = 2, 95%CI 0.45–13.55). However, it supported the association genital-anal with an SIR of 13.97 (95%CI 5.03–27.39). The association was additionally supported by the male anal-genital SIR of 60.24 (95%CI 5.68–172.66).
The oncogenic potential of infective agents, such as HPV, can be shown by demonstrating viral DNA or specific antibodies in the human body.2, 4 Epidemiological studies, lacking data on past infections, cannot establish the causative agent. However, because large numbers of individuals can be studied, etiological hypotheses can be proposed for later confirmation with molecular techniques. The present results were consistent with some previous findings, particularly in showing the association of the first and second cancers at known HPV-related sites: anus, cervix and male and female genitals. These associations were strong, and they were often confirmed in the two-way analysis among men and women. Oral cancers were associated, equally strongly, with these sites, consistent with the proposed HPV etiology.9–12 However, some contribution by smoking is likely in this association because we were not able to control for smoking, and because smoking appears to be common among women who present with cervical cancer. This may be due to the carcinogenic effects of smoking on the cervix or to a life-style characterized by smoking and sexual promiscuity.20–22
The most interesting novel observation was the association of rectal cancer with HPV-related sites. In men, rectal cancer was followed by an excess of oral and anal cancers, the latter with a borderline significance. Skin cancer was also increased, but because this is not a verified HPV site, we discuss skin cancers later. In men, an increased risk of rectal cancer was observed after oral, anal and genital cancers. Rectal tissue will obviously receive a radiation dose if anal cancer is treated with X-rays. However, the effects of radiotherapy in causing solid tumors are usually observed some 10 years or more after treatment.14, 23 The risk for second rectal cancer after anal cancer was also observable in the early period of follow-up, making the sole contribution by radiotherapy unlikely. Treatment for the first cancer may also cause impairment of the patient's immune or other defence mechanisms against latent infections. Such an effect may influence the risk of second cancer soon after treatment, as has been observed among kidney transplantation patients after immunosuppressive therapy.24 We have no possibility of controling for such effects. In women, rectal cancer was followed by an increased risk of genital cancers. This finding could be potentially biased by the effects of X-rays. However, again the increased risk was observed throughout the follow-up period. Cervical cancer patients had an increased risk of rectal cancer, in agreement with an earlier report.23 Taken together, these results, in both men and women, showed a clear association of rectal cancers and HPV-related cancers, which apparently cannot be explained by the effects of treatment. HPV causes both cervical adenocarcinomas and squamous cell carcinomas, providing a precedent for the possible adenocarcinoma formation in the rectum.2, 6 An epidemiological argument supporting the association is the lack of known shared risk factors between the rectum and the HPV-related sites, making bias an unlikely explanation. However, whether HPV can be incriminated in rectal cancer requires the demonstration of a prior infection in patients.
The association of skin cancer with the HPV-related sites was also suggestive, although this was not as convincing as the above associations. In men, skin cancer was increased after oral and rectal cancer. In women, skin cancer was increased after oral and anal cancer. However, the SIRs were over unity at many other HPV-related sites, but due to the small number of cases, the increases were not significant. In earlier follow-up studies on skin cancer, increases in oral cancer have been observed.25, 26 In family studies an association has been observed between maternal cervical cancer and daughter's skin cancer, and vice versa between maternal skin cancer and daughter's cervical cancer.21, 27 However, such data do not help to distinguish vulnerability to HPV infection due to promiscuous life-style, or genetic factors, including depressed immune function.21, 28
In summary, we showed consistent increases in second HPV-related cancers, including oral cancers, when the first cancer was an HPV-related cancer. As a novel finding, an association of rectal cancer with the HVP-related cancers was shown. New evidence on the association of skin cancer with the HPV-related cancers was also provided. As an epidemiological study, the associations were strong and often supported by a number of comparisons. They could not be explained by bias, and there was no evidence on long-term treatment related effects that would modify the conclusions. However, whether the findings on the rectum and skin are due to HPV or other infections, transient or inherited depressed immune function or other constitutional factors remains to be established.
- 2IARC. Human papillomaviruses. Lyon: IARC, 1995.
- 19Statistical methods in cancer research. Lyon: IARC, 1994., , .