Second primary neoplasms in 633,964 cancer patients in Sweden, 1958–1996
The Swedish Family-Cancer Database was used to analyze concordant (same site) and discordant (different site) second primary neoplasms in 633,964 cancer patients diagnosed from 1958 to 1996. Cases of second malignant neoplasms were extracted from the Database if the diagnosis date of the first and second cancer differed by at least 1 month. The expected numbers of cancers were obtained by applying site-, sex-, age-, period-, residence- and socioeconomic level-specific rates in the corresponding population in the Database to the appropriate person-years at risk. The standardized incidence ratio (SIRs) of a second cancer was taken to be the ratio of observed to expected numbers of second cancers. Of all cancers, 8.5% were subsequent neoplasms (8.4% for males and 8.7% for females). SIRs for both concordant and discordant subsequent cancer were elevated in patients with cancer of the upper aerodigestive tract, colon, nose, breast, other female genitals, testis, kidney, urinary, bladder, skin, nervous system, endocrine, bone, connective tissue, melanoma, lymphoma and leukemia. The risks at some concordant sites, such as nose, squamous cell skin, bone and connective tissue in both sexes, breast in males and upper aerodigestive tract and leukemia in females, were very high (>10). At discordant sites, SIRs were less than 2 but significantly increased after all but gastric and prostatic cancer. Compared with the general population, cancer patients were at a modestly increased risk for new primary cancer after cancers at many sites, calling for attention in treatment, management and prevention. © 2001 Wiley-Liss, Inc.
It has been recognized that, owing to shared risk factors, some cancers tend to cluster more frequently than expected, for instance, tobacco smoking in cancers of the lung, head and neck, pancreas, bladder and kidney,1–9 dietary or endocrine factors in gynecological cancers,10, 11 ultraviolet light in melanoma and squamous cell carcinoma of the skin12–14 and viral agents in cervical and anogenital cancers, Karposi's sarcoma and lymphomas.15
There is evidence that second primary malignancies may be associated with a potentially carcinogenic treatment of the initial cancer, such as radiation therapy or chemotherapy.16 The identification of an increased risk of second cancers due to treatment may result in changed treatment protocols and more intensive surveillance of patients. Hereditary predisposition may also play a role in the coexistence of multiple cancers, which form some rare familial cancer syndromes such as hereditary non-polyposis colorectal cancer (HNPCC), Li-Fraumeni syndrome, breast-ovarian cancer syndrome and multiple endocrine neoplasia (MEN).17–19 It is therefore important for both etiological and clinical purposes to study multiple primary cancers.
A considerable number of large population-based studies have been conducted on the risks of second primary cancers in Sweden, but all focused on determination of the specific patterns of second primary cancers for which a cancer patient may be at increased risk. In this nationwide population-based study, we systematically analyzed the risks of developing a second primary cancer among 633,964 cancer patients diagnosed with a specific cancer during 1958 to 1996 in Sweden. Because of a great number of combinations (35 × 35 = 1225) of first and second cancer sites, we here only reported the risk of second primary cancers at concordant and all discordant sites after initial primary cancer at a specific site.
SUBJECTS AND METHODS
The Swedish Family-Cancer Database
The Swedish Family-Cancer Database, updated in 1999, was formed from the Second Generation Register maintained by Statistics Sweden and linked by the individually unique national registration number to the Swedish Cancer Register at the National Board of Health and Welfare. The Database includes all persons born in Sweden after 1934 with their parents, totaling over 9.6 million individuals.20, 21 Since 1958, all new cases of cancer in Sweden have been reported to the Swedish Cancer Register. An almost 100% coverage has been achieved by compulsory reporting from clinicians who diagnose a neoplasm and the pathologists/cytologists who must report separately any diagnosis of cancer made on pathological and cytological specimens. The Database has an almost complete follow-up of the registered cancer patients, and it provides a unique possibility to quantify the short- and long-term risks of developing a second primary cancer among patients with an initial primary cancer.
The site of cancer is registered based on a four-digit diagnostic code according to the 7th revision of the International Classification of Diseases (ICD-7). The following ICD-7 codes were pooled: “upper aerodigestive tract” cancer codes 161 (larynx) and 140 to 148 (lip, mouth, pharynx), except for code 142 (salivary glands), “lymphoma” codes 200 (non-Hodgkin's lymphoma), 201 (Hodgkin's disease) and 202 (reticulosis), and “leukemia” codes 204 to 207 (leukemias), 208 (polycythemia vera) and 209 (myelofibrosis). Rectal cancer, ICD-7 code 154, was separated for anus (squamous cell carcinoma, 154.1) and mucosal rectum (154.0). For skin cancer, ICD-7 code 190 was used for melanoma and 191 for squamous cell carcinoma of the skin. Basal cell carcinoma of the skin is not registered in the Cancer Registry.
All patients with a first malignancy diagnosed between January 1, 1958 and December 31, 1996 were retrieved from the Database, a total of 633,964 cases. The 35 subcohorts (first cancer by site) were matched with the same register to identify second primary cancers. Cases of second malignant neoplasms were extracted from the Database if the diagnosis dates of the first and second cancers differed by at least 1 month and by different topology when available (using the four-digit ICD code for concordant sites). Between 1958 and 1996, 26,582 and 27,502 cancer patients developed a second primary cancer in males and females, respectively.
Person-years at risk were accumulated for each subject beginning with the date of diagnosis of the first primary cancer and ending with the date of diagnosis of a second primary cancer, date of death, date of emigration or December 31, 1996, whichever came first. For each subcohort, person-years at risk were classified by sex, 5-year age group, 5-year period (1958 to 1962, 1963 to 1967, 1968 to 1972, 1973 to 1977, 1978 to 1982, 1983 to 1987, 1988 to 1992 and 1993 to 1996), two-category residential area (three counties of large cities vs. the rest), four-category socioeconomic level (farmer, worker, professional and the rest) and time since entry to the cohort (<1, 1 to 9 and <9 years). More than 9 years after the first diagnosis was selected as a cutoff point to explore the potential radiotherapy-induced risk for second primary cancers.22
The expected numbers of cancers were obtained by assuming that these persons experienced the same cancer incidence as prevailed in the corresponding general population in the Database and by applying site-, sex-, age-, period-, residence- and socioeconomic level-specific rates to the appropriate person-years at risk. The standardized incidence ratio (SIR) of a second cancer was taken to be the ratio of observed (O) to expected (E) numbers of second cancers. Confidence intervals (95% CI) for SIRs were calculated assuming that the cases followed a Poisson distribution.23
Characteristics of patients with an initial primary cancer
For each cohort, separated by sex, Table I gives the size at start of the follow-up, total person-years at risk, median age at first diagnosis, median length of follow-up, number and percent of second cancers and SIRs for a second primary cancer. There were 316,642 men and 317,322 women with an initial primary cancer. In male patients, the median age at first diagnosis was 67 years, and the median follow-up time was 2 years. A total of 1,539,069 person-years at risk were counted, and a second primary cancer was found in 26,582 patients (8.4%), giving an overall SIR of 1.3 (95% CI = 1.2 to 1.3). In female cancer patients, the median age at first diagnosis, median follow-up time and person-years at risk were 60 years, 3 years and 2,238,423, respectively. Of the female cancer patients, 27,502 (8.7%) had a subsequent primary cancer, giving an overall SIR of 1.6 (95% CI = 1.5 to 1.6).
Table I. Characteristics of Persons With An Initial Primary Cancer1
|Upper aerodigestive tract||10,942||79,618||63||4||1811||16.6||1.8||1.7–1.9||3405||23,043||63||4||395||11.6||2.0||1.8–2.2|
The highest risks for second cancers were seen in male patients with an initial cancer of the squamous cell skin (2.3), small intestine (2.0), upper aerodigestive tract (1.8) and testis (1.8) and lymphoma (1.8), whereas the highest SIRs were found in female patients with an initial cancer of the upper aerodigestive tract (2.0), squamous cell skin (1.9), bone (1.9) and breast (1.8). Decreased SIRs were found in patients with cancer of the pancrease (0.7 for both sexes), stomach (0.8 for male and 0.9 for female), prostate (0.8) and other sites (0.7 for male and 1.0 for female). At sites of salivary glands, breast, skin (melanoma) and thyroid, the median age at first diagnosis in female patients was at least 4 years younger than that in male patients, whereas at sites of non-thyroid endocrine glands and bone, female patients had at least 4 years older median age at first diagnosis than male patients.
Concordant subsequent primary cancer
Table II includes the sites at which the mean time of follow-up was more than 2 years and there were at least five concordant second primary cancers in the cohort. SIRs of concordant second events after an initial primary cancer are shown by site, sex and follow-up interval. Through all follow-up periods, there was a decreased risk of gastric and rectal cancer in each sex and of prostatic, cervical, endometrial and ovarian cancer in either sex. Male breast cancer patients ran a risk of 111.7 (95% CI = 10.5 to 410.9) of developing a contralateral breast cancer more than 9 years after the first diagnosis. When both sexes were combined, compared with the general population, patients had an over 5-fold risk of developing a concordant cancer after cancer of the upper aerodigestive tract, nose, skin and connective tissues and after melanoma and leukemia through all follow-up periods. Compared with the period 1 to 9 years of follow-up, the period 10 to 38 years of follow-up showed an increase in SIRs at colon, nose, female genitals, testis, kidney, urinary bladder, endocrine and bone and after lymphoma and leukemia. Among those sites with an increased SIR, the highest SIR was seen within the first year of diagnosis, except for nose, bone and leukemia.
Table II. Risk for Subsequent Concordant Primary Cancer in Patients With An Initial Cancer1
|Upper aerodigestive tract||138||6.5||5.4–7.6||50||5.0||3.7–6.6||52||29.9||22.3–39.2||15||19.5||10.9–32.2||61||23.5||18.0–30.2||129||6.3||5.3–7.5||65||6.0||4.7–7.7|
Discordant subsequent primary cancer
Table III reports the same initial cancer sites as Table II and shows the SIRs for all discordant subsequent primary malignancies combined after a specific primary cancer. During the first 9 years after the first diagnosis, there was an overall risk, ranging from 1.2 to 1.8 in male patients and 1.2 to 2.0 in females patients, for discordant cancers; gastric cancer was an exception (SIR = 1.0). All the elevated risks were statistically significant, with the exceptions of male breast cancer (95% CI = 0.9 to 1.7) and female nasal cancer (95% CI = 0.9 to 2.2). When both sexes were combined, the highest risk for a subsequent primary cancer was seen <1 year of follow-up after an initial primary at each site. Compared with the period 1 to 9 years of follow-up, the period 10 to 38 years of follow-up showed a very slight or moderate increase in SIRs among patients with an initial cancer of the rectum, nose, female breast, cervix, endometrium, ovary, testis, thyroid and endocrine glands and lymphoma, no change in SIR among patient with an initial cancer of the colon, female genitals, kidney, urinary bladder and bone and melanoma as well as myeloma and a slight decrease in SIR for patients with an initial cancer at the remaining sites. Compared with the general population, a significant decreased risk for discordant subsequent cancer was observed in gastric cancer patients ≥1 year after first diagnosis and in male nervous system cancer patients >9 years after first diagnosis.
Table III. Risk For Subsequent Discordant Primary Cancer In Patients With An Initial Cancer1
|Upper aerodigestive tract||1091||1.7||1.6–1.8||532||1.5||1.4–1.6||259||2.0||1.7–2.2||69||1.1||0.9–1.4||200||2.4||2.1–2.7||1150||1.7||1.6–1.8||601||1.4||1.3–1.6|
Second primary tumors arise because of either inherited or acquired mutations or deficiencies. Second primary tumors may develop soon or very late after treatment of the first primary tumor and may reflect underlying genetic or immunologic defects in the patient, treatment-related damage or environmental exposure to carcinogens. With the increasing success of modern chemotherapy and radiotherapy in achieving long-term remissions in many patients, second primary tumors are a rapidly increasing disease category.
It is sometimes impossible to distinguish second tumors as independent primaries. The Swedish Cancer Registry has clear instructions about the reporting of multiple primary malignancies, and a re-evaluation of 209 multiple primary tumors found 98% of second malignancies to be correctly classified.24 The incidence of second malignancies may depend on factors such as age at diagnosis. In the present study we had no possibility of examining the determinants of second malignancies in detail, but for breast and colon cancer and for melanoma, such studies have been published from this Database (see refs. 25 and 26 and the references therin).
Second cancers in the database
Here we used the 1999 update of the nationwide Swedish Family-Cancer Database to analyze the occurrence of second primary cancers systematically in 316,642 males and 317,322 females diagnosed with an initial primary cancer compared with site-, sex-, age-, period-, residence- and socioeconomic level-specific incidence rates of all men and all women in the Database.
In the present study, there are three important technical points. One is that second cancers were included if the first and second diagnosis differed by at least 1 month and topology. Topology involved a different anatomic site, if specified by the fourth digit of the ICD-7 code. Another is that we divided the follow-up time into three periods, <1, 1 to 9 and >9 years. The period >9 years of follow-up may reveal the effects due to radiotherapy and chemotherapy,10, 11 prompting comparison of the SIRs between the two periods. The period shortly after the first diagnosis may reveal the effects of intense medical surveillance. The third point is the estimation of SIRs based on adjustment for residence and socioeconomic status, which may be the potential confounding factors and which have been ignored by most studies on second cancers.
Overall, compared with the general population, male and female patients were at risk of 1.3 and 1.6 for developing a second cancer, respectively. The SIRs are somewhat higher than those observed in a Finish study (1.00 for male patients and 1.25 for female patients).27 This may be due to the different baseline incidence rates (in the present study, we adjusted for residence and socioeconomic class) and the different diagnosis periods (in the present study, data included patients diagnosed from 1958 to 1996). Of the male second primary cancers, 51% followed cancer of the prostate, urinary bladder, colon and skin (non-melanoma), whereas 53% of female subsequent cancers occurred after cancer of the breast, endometrium and cervix. Due to poor survival rate, less than 4% of the patients with cancer of the esophagus, small intestine, liver, pancreas and lung had a second primary tumor in either sex.
Second cancers at concordant sites
At all common cancer sites, both sexes showed a similar pattern (increasing or decreasing) for risk of developing a concordant subsequent cancer, although at some sites, such as upper aerodigestive tract, nose, breast, urinary bladder and thyroid, males differs from females in SIR due to the difference in the background incidence in the general population. A low background rate in one gender conveyed a high risk for the subsequent cancer compared with the gender of the high background rate. There was an obvious deficit in SIR for concordant subsequent cancers after gastric and rectal cancer in both sexes and prostatic, endometrial, cervical and ovarian cancer in males or females. There are at least two explanations for the deficit of second cancer in these patients. One is that all or part of the organ is removed at the operation for the first cancer; another is that in cancers of poor prognosis, reduced surveillance by physicians leads to the underregistration of subsequent cancers.
Compared with the general population, patients ran an increased risk of developing a concordant subsequent cancer at the sites where organs or parts of organs remained after treatment for the first cancer. The SIR was over 10 for second cancer at the concordant site after nasal, squamous cell skin, bone and connective tissue cancer in both sexes, male breast cancer and female upper aerodigestive tract cancer and leukemia. Compared with the 1 to 9 year period of follow-up, the >9 years period of follow-up showed an increasing risk for a second cancer in patients with cancer of the colon, nose, female genitals, testis, kidney, urinary bladder, non-thyroid endocrine glands and bone, and for lymphoma and leukemia when both sexes were combined. Radiotherapy is known to increase the incidence of subsequent primary cancers, and there is a strong link between chemotherapy and leukemia and bladder cancer.22, 28–33 Our results agree with the patterns for radiation-induced subsequent cancer and suggest that radiotherapy, possibly combined with chemotherapy, may be a contributor to the excess risk for these cancers.
It is noteworthy that there was a more than 3-fold excess of concordant subsequent cancer during the first 9 years after the first diagnosis among patients with cancer of the upper aerodigestive tract, nose, female genitals, testis, melanoma, skin, nervous system, bone and connective tissue, as well as lymphoma and leukemia. This, of course, could partly be explained by the intensive clinical surveillance preceding or following treatment. However, because almost all cancers are both histologically and clinically verified in the Swedish Cancer Registry, false diagnosis is unlikely. Instead, diagnosis will be rendered at an earlier time period. The results also suggest that the shared exposure to environmental or inherited risk factors leads to multiple cancers at a specific site. There are only a few population-based reports available on the role of family history of cancer in the development of second cancer, including breast and multiple skin cancer.4, 20, 34–36
Second cancers at discordant sites
For second cancers at discordant sites, the <1 year period of follow-up showed the highest risk for a subsequent primary cancer regardless of the first cancer sites, suggesting an effect of intensive medical surveillance. Compared with to the general population, cancer patients with an initial cancer at most sites ran a 10 to 50% excessive risk for a new primary cancer at discordant sites, indicating that second cancers are not a major public health problem at a population level. The SIRs changed only marginally between the period 1 to 9 year and >9 years since the first diagnosis, suggesting that the observed excessive risk could not be completely accounted for by the effects of treatment but rather reveals a role for shared risk factors in the etiology of multiple primary cancers.
In general, apart from the sites with a single organ, our data show that cancer patients are at higher risk of developing a second primary malignancy at concordant site than at discordant sites. The risk at some concordant sites, such as nose, squamous cell skin, bone and connective tissue in both sexes, breast and upper aerodigestive tract cancer in males and leukemia in females, was very high (>10). At discordant sites, SIRs were less than 2 but increased after all but gastric cancer. These data provide evidence that cancer at discordant sites (tissues) could also share some risk factors (environmental or/and hereditary) that lead to multiple cancers. Further studies concerning the contributors to the multiple malignancies, at concordant and discordant sites, are needed for a better understanding of the etiology of second primary neoplasms.