Familial risk of cancer: Data for clinical counseling and cancer genetics
Familial risks for cancer are important for clinical counseling and understanding cancer etiology. Medically verified data on familial risks have not been available for all types of cancer. The nationwide Swedish Family-Cancer Database includes all Swedes born in 1932 and later (0–to 68-year-old offspring) with their parents, totaling over 10.2 million individuals. Cancer cases were retrieved from the Swedish Cancer Registry up to year 2000. Standardized incidence ratios (SIR) and 95% confidence limits (CI) were calculated for age-specific familial risk in offspring by an exact proband status. The familial risks for offspring cancer were increased at 24/25 sites from concordant cancer in only the parent, at 20/21 sites from a sibling proband and at 12/12 sites from a parent and sibling proband. The highest SIRs by parent were for Hodgkin's disease (4.88) and testicular (4.26), non-medullary thyroid (3.26), ovarian (3.15) and esophageal (3.14) cancer and for multiple myeloma (3.33). When a sibling was affected, even prostate, renal, squamous cell skin, endocrine, gastric and lung cancer and leukemia showed SIRs in excess of 3.00. The highest cumulative risks were found for familial breast (5.5%) and prostate (4.2%) cancers. We identified reliable familial risks for 24 common neoplasms, most of which lack guidelines for clinical counseling or action level. If, for example, a familial SIR of 2.2 would be use as an action level, counseling would be needed for most cancers at some diagnostic age groups. The present data provide the basis for clinical counseling. © 2003 Wiley-Liss, Inc.
Familial cancer is an avenue to the understanding of cancer etiology, giving the first indication of the possible involvement of heritable genes.1, 2, 3, 4, 5 Familial clustering is also the basis for clinical decisions, counseling and genetic testing for cancer risk.6 The main problem in the published literature on familial risks is the possible inaccuracy of data on cancer in family members, who had died a long time before the study. There is ample literature illustrating the problem, which has been least for breast cancer, intermediary for prostate cancer and melanoma and worst for internal neoplasms, for which the accuracy of reporting may be <50%.12, 13, 14, 15, 16, 17, 18 This level of inaccuracy may cause severe bias in the derived risk estimates, with a tendency to report exaggerated risks, in the most commonly used case-control studies.19, 20, 21, 22 A failure to consider the level of false reporting is the basis of common misquotations on the proportion of familial cancer among all cases; for example, an authoritative body representing the World Health Organization classification of tumors recently stated on gastric carcinoma that ‘familial clustering occurs in 12–25% with a dominant inheritance pattern’; a study with medically verified cases of gastric cancer observed only a clustering of 1.9% and an unclear inheritance pattern.20, 23
Cancer syndromes, such as hereditary nonpolyposis colorectal cancer (HNPCC) or BRCA1/2-related breast and ovarian cancer, manifest a certain pattern of tumors, which are used clinically as guidance for further action.8 In prostate cancer, even though susceptibility genes have remained elusive, recommendations on clinical follow-up are given depending on diagnostic ages of known cases in family members.6, 8, 24 For most other cancers, however, no guidelines have been elaborated for the cause of action when a patient is being advised on the likelihood on his or her risk based on cancers in family members. A major reason for the lack of guidelines on familial cancer at large is likely to be the sparseness of reliable data; the biased risk estimates, discussed above, may not be an appropriate cause for clinical counseling. In our study, we resort to a unique Swedish dataset, the nation-wide Family-Cancer Database that we have formed by combining a national dataset on families with the Swedish Cancer Registry. All cancer data are verified medically and both cancer and family data have a practically complete national coverage, guarding against bias. The present results are derived from the most recent update of the Family-Cancer Database, including more than 1 million tumors from years 1958–2000. We present age-group specific familial risks in all common cancers and discuss their implications for clinical counseling.
MATERIAL AND METHODS
Statistics Sweden maintains a “Multigeneration Register” where children, offspring, born in Sweden in 1932 and later are registered with their parents (those pleading parenthood at birth) as families.25 Information on the Database is also available at the Nature Genetics website as ‘Supplementary information’ to reference.26 The data on families and cancers have a complete coverage, baring some groups of deceased offspring, which affect those born in the 1930s and who died before 1991. Although this small group of offspring with missing links to parents has negligible effect on the estimates of familial risk,27 we limited the present study to offspring diagnosed between years 1991–2000, to eliminate any possibility of bias. This Register was linked by the individually unique national registration number to the Cancer Registry from years 1958–2000. Because of some inaccuracies in vital status determination in the first years of cancer registration, however, parental cancer in the present study was diagnosed between years 1961–2000. Cancer registration is considered to be close to 100% currently.28 A 4-digit diagnostic code according to the 7th revision of the International Classification of Diseases (ICD-7) and subsequent ICD classifications are available. Additional linkage was carried out to the national census data to obtain socio-economic background data and to death notifications for vital status determination. The linkages were carried out at Statistics Sweden who delivered the final matched records in an unidentified form.
Standardized incidence ratios (SIRs) were used to measure the cancer risks for offspring according to occurrence of cancers in their families. SIRs were calculated for offspring whose parent only, sibling only or parent and sibling had the same, concordant cancer, i.e., using parents only, sibling only or parent and sibling as probands. Follow-up was started for each offspring at birth, immigration or January 1, 1991, whichever came latest. Follow-up was terminated on diagnosis of first cancer, death, emigration, or the closing date of the study, December 31, 2000. When more than 2 affected offspring were found in any family, they were counted as independent event. This caused statistical dependence in families were 2 or more siblings were affected, as discussed in the following paragraph.
Parents' ages were not limited but offspring were 0–68 years of age. Data were shown for ICD-7 sites only if at least 3 familial pairs or 2 triplets were identified for the particular proband status. SIRs were calculated as the ratio of observed (O) to expected (E) number of cases. The expected numbers were calculated from 5-year-age-, gender-, tumor type-, period- (5 year bands), socioeconomic status- (6 groups) and residential area- (3 groups) specific standard incidence rates for all offspring lacking a family history.29 The sporadic reference population probably contained smaller families than the familial population; however, based on our previous study, family size does not influence familial risk.30 For each of the 25 types of cancer analyzed, we used 3 reference datasets for the calculation of expected numbers: (1) offspring with no family history in parents; (2) offspring with no family history in siblings; and (3) offspring with no family history in parents nor siblings. Offspring risks were shown separately for early onset cancers (diagnosed before age 50 years) and for those limited by probands' diagnostic age (sibling before age 50 years or parent before 60 years). Confidence intervals (95% CI) were calculated assuming a Poisson distribution.29 Risks for siblings were calculated using the cohort method, described elsewhere.31 In this method, families with multiple affected individuals are ascertained at multiple times and they are not independent, leading to too narrow CIs (approximately by a factor of 1.432); no correction was done in our present study. Cancer incidence was calculated based on the Family-Cancer Database and the rates were adjusted to the standard European population. Cumulative incidence was calculated for offspring up to age 68 years.
The year 2002 update of the Swedish Family-Cancer Database included 754,165 first invasive parental (diagnosed at any age in years 1961–2000) and 112,216 offspring (diagnosed between ages 0–68 years in years 1991–2000) cancers. Familial risks were calculated for 89,567 offspring when only parent was affected by a concordant cancer for sites at which >3 familial pairs were found (Table I). A total of 4,938 familial cancers were found, with an overall SIR of 2.02. Among offspring diagnosed at any age, all the site-specific familial risks were significantly increased, except that for connective tissue. Hodgkin's disease showed the highest SIRs of 4.88, followed by testicular (4.26) and non-medullary thyroid cancer (3.26). Esophageal and ovarian cancer and multiple myeloma had SIRs in excess of 3.00. Among common cancers the SIRs were increased for female breast (1.84), prostate (2.45) and colorectal adenocarcinoma (1.86), and the number of familial pairs ranged between 681 and 1,779 for each. Restricting the age of onset for the offspring to <50 years or that of probands to <60 years increased the SIRs at many sites to an approximately equal extent, although, because of smaller numbers of cases, only the increases at the colorectum, breast and prostate were significant (non-overlapping 95% CI). When the diagnostic ages of both offspring and parents were restricted, the SIR of Hodgkin's disease was 6.71, and that of endometrial, ovarian and prostate cancer over 5.00; even the SIR of colorectal adenocarcinoma was 4.51.
Table I. SIR for Cancer in Offspring by Parental Concordant Cancer at Different Diagnostic Ages1
|Upper aerodigestive tract||2104||39||1.72||1.22||2.35||12||1.81||0.93||3.17||9||1.94||0.88||3.69||4||1.95||0.51||5.04|
|Skin, squamous cell||1898||77||2.52||1.99||3.15||18||2.61||1.54||4.13||3||1.46||0.28||4.33||1||1.12||0.00||6.40|
|Thyroid gland, nonmedullary||1208||12||3.26||1.67||5.71||11||4.49||2.23||8.06||8||4.11||1.76||8.14||8||4.94||2.11||9.79|
|Myeloma||901||23||3.33||2.11||5.00||4||2.66||0.69||6.88||2||2.50||0.24||9.20||0|| || || |
Table I also shows, as the first column, ‘all cases in offspring,’ which can be used to calculate the proportion of offspring with an affected parent, the familial proportion. Familial cases constituted 15.3% of all prostate cancers (922/5,993), which was the largest proportion by far. The proportion was 10.1% for colorectal and 8.5% for breast cancer but only 0.4% for connective tissue and 0.5% for testicular tumors.
Table II shows familial risks among siblings for sites with at least 3 affected pairs. All sites showed a significant effect, except for upper aerodigestive tract cancer; the effect for pancreatic cancer was of borderline significance. Testicular cancer showed the highest SIR of 9.28, followed by Hodgkin's disease (5.94), kidney (4.74), prostate (4.46), and ovarian cancer (4.25). When the age of onset of the proband was limited to <50 years, the SIR for prostate cancer was 10.58 and that for testicular cancer 9.48. When both siblings were diagnosed before age 50 years, the SIR of prostate cancer reached 149.64, and SIRs of stomach, lung, endometrial, ovarian and testicular cancers exceeded 10.00.
Table II. SIR for Cancer in Offspring by Sibling Concordant Cancer at Different Diagnostic Ages1
|Upper aerodigestive tract||6||1.41||0.51||3.10||0|| || || ||1||0.64||0.00||3.67||0|| || || |
|Pancreas||6||2.75||0.99||6.04||0|| || || ||1||2.64||0.00||15.15||0|| || || |
|Lung||92||3.13||2.53||3.84||0|| || || ||28||4.64||3.08||6.71||15||10.49||5.86||17.35|
|Prostate||189||4.46||3.85||5.15||0|| || || ||9||10.58||4.80||20.18||4||149.64||38.92||386.92|
|Skin, squamous cell||12||3.63||1.87||6.37||1||2.55||0.00||14.63||3||2.68||0.51||7.94||2||5.90||0.56||21.68|
|Thyroid gland, nonmedullary||6||3.89||1.40||8.53||3||3.61||0.68||10.69||5||4.00||1.26||9.42||3||3.97||0.75||11.75|
The third proband status, both a parent and a sibling affected, resulted in high SIR for many neoplasms (Table III). Only 215 familial triplets were diagnosed for any of the cancers listed. Considering any age, endocrine gland and ovarian tumors had the highest SIRs of over 30.00; even the SIR for stomach (12.66) and endometrial (12.54) and squamous cell skin cancer (17.24) were very high. Young age of onset was a risk factor: SIR for colorectal adenocarcinoma was 15.49 when a parent was diagnosed before age 60 years and that for prostate cancer was 29.80 when the sibling was diagnosed before age 50 years. The SIRs for colorectal, ovarian and nervous system cancers and melanoma were very high when both probands were diagnosed at a young age.
Table III. SIR for Cancer in Offspring by Both Parental and Sibling Concordant Cancer at Different Diagnostic Ages1
|Stomach||2||12.66||1.19||46.55||0|| || || ||0|| || || ||0|| || || |
|Lung||8||5.06||2.16||10.02||2||9.51||0.90||34.99||0|| || || ||0|| || || |
|Prostate||55||8.62||6.49||11.23||4||15.08||3.92||38.99||6||29.80||10.73||65.30||0|| || || |
|Urinary bladder||3||8.16||1.54||24.16||0|| || || ||1||8.19||0.00||46.97||0|| || || |
|Skin, squamous cell||2||17.24||1.63||63.41||0|| || || ||0|| || || ||0|| || || |
|Endocrine glands||3||35.91||6.77||106.30||0|| || || ||3||56.92||10.73||168.50||0|| || || |
In Table IV we show absolute risk of cancer in the offspring population up to age 68 years. Familial risks are considered for a parental proband status. Breast and prostate cancers showed the highest incidence rates for familial and sporadic cases; connective tissue tumors had the lowest rate. The cumulative incidence of sporadic breast cancer was 3.4% and that for daughters of affected mothers 5.5%. For prostate cancer, the risk was 2.1% for sporadic and 4.2% for familial cases. For colorectal adenocarcinomas, the risks were 1.6 and 2.6%, respectively.
Table IV. Absolute Risk in Familial and Sporadic Cancer Among Offspring
|Upper aerodigestive tract||39||6.0||0.5||2065||4.5||0.4|
|Skin, squamous cell||77||10.1||0.9||1821||4.8||0.4|
|Thyroid gland, nonmedullary||12||6.4||0.5||1196||1.9||0.1|
Familial clustering of cancer has been traditionally studied in the clinical setting where probands and their multiple affected relatives have been identified.33, 34 This approach has been very productive also in terms of understanding cancer genetics. Many forms of cancer in which a single gene poses a high risk have been identified. The disadvantages include difficulties in obtaining large numbers of cases and in securing unbiased risk estimates. These problems can be overcome in population datasets where cancer data on cases and probands are medically verified, including the Utah Population Database and the Icelandic population records.32, 35 The largest population-based dataset on familial cancer is the Swedish Family-Cancer Database, covering 10.2 million people and 1.1 million tumors in its most recent updated form from year 2002, used in our present study. It has been the main source of recent global literature on medically verified familial cancers. Although more that 130 publications have emanated from the Database, it has never been used in a comprehensive analysis on familial cancer with a clinical perspective. The previous study on all common types of cancer was based on a younger offspring population, it had low statistical power compared to the present study and no age-specific analysis could be carried out.36 Many of these earlier studies have had a genetic perspective and they have discussed familial risks in site-specific cancers much more extensively than we can do here.
Our present study showed familial risks for all cancer sites, except for the connective tissue. Even though some of the risks were high, the absolute risk, which is more relevant than the relative risk in a counseling situation, was still small even for common cancers in this 0–68 year old population. The highest cumulative risks were for breast (5.5%) and prostate (4.2%) cancer when a parent was affected by a concordant cancer. In an older population the cumulative risks would increase but the relative familial risks decrease; however, the overall effect would be a higher cumulative risk. Diagnosis of cancer in a family member, particularly in a sibling, is likely to alert the remaining relatives to search for symptoms in themselves and to seek for medical advice. Thus, an apparent excess, lead time bias, would be noted in cancers for which screening methods are available, such as prostate, breast, colorectal and skin cancers. In the Swedish Cancer Registry, however, an overwhelming proportion of diagnosis are histologically verified and false diagnosis is thus unlikely.
Familial aggregation of cancer may be due to environmental factors shared by family members or due to shared genes.37 We have tried previously to estimate the degree of environmental sharing by comparing cancer risks between spouses but spouse concordance, not generally exceeding an SIR of 1.4, can only be noted for cancers with known strong environmental risk factors: lung and genital cancers and early onset gastric cancer and melanoma.37, 38 Thus, for most other sites, heritability is likely to be the main contributor. Environmental factors are probably the main contributor to the familial aggregation of cervical, lung and upper aerodigestive tract cancers, and minor contributor to familial risks for melanoma and squamous cell skin cancer.39, 40 For lung cancer, a sibling risk of 10.49 when both siblings were diagnosed before age 50 years, seems very high to be explained solely by smoking. A clinical counseling on tobacco-related and sexually transmitted cancers is concerned mainly on the avoidance of risk factors rather than family history, however, and efficient screening methods are available for cervical cancer.41 In the case of melanoma and squamous cell skin cancer, counseling has to consider both; an affected sibling with one of these skin neoplasms signaled a 3-fold risk, and if additionally a parent was affected, the risk was 10-fold or more. Such a high risk is likely to be due to a genetic condition, such as CDKN2A mutation in melanoma; patients with mutation often present with multiple melanomas.42, 43, 44
There are no commonly accepted clinical counseling guidelines or action plans for familial cancer at large. For familial breast and ovarian cancer guidelines have been developed for genetic testing for BRCA1/2, and these consider age of onset and gender of the affected family members and their tumors.9, 45 Newly detected low-penetrance genes, such as CHEK2, however, also need to be considered.46 Similar guidelines have been developed for hereditary colorectal cancer, also encompassing endometrial cancer.9, 47, 48 When a hereditary syndrome has been excluded, the American Cancer Society Guidelines recommend colonoscopy every 5–10 years if any first-degree relative is diagnosed with colorectal cancer or adenomatous polyps before age 60 years.8 According to the present Table I, the risk of colorectal adenocarcinoma was 3.04 if a parent was diagnosed before age 60 years, and according to Table II, the risk to a sibling was 2.87 if another sibling was diagnosed before age 69 years. Because the American Cancer Society Guidelines include any first degree relatives and encompass even polyps, we can assume that it would correspond in the present study to an approximate familial risk of 2.5 for colorectal cancer from one type of proband. This calculated familial risk includes hereditary colorectal cancer syndromes, however, which probably account for some 20% of familial colorectal cancer in this Database.49, 50 Thus the SIR for nonhereditary familial colorectal cancers would be 2.2 at the estimated action level proposed by the American Cancer Society Guidelines for this relatively common type of cancer. Although many types of data need to be considered for an action level of familial cancer, the data in Tables I and II show that familial SIRs exceeded 2.2 for most cancers at least at some age groups. These data would urgently call for site-specific or more uniform guidelines on a clinical counseling and action plan for familial cancer.
The proportion of familial cancer of all cancer depends on many factors, even when the ascertainment of families and cancer cases is complete, including age of the population and any age truncations, family size and number of generations. The present results apply to a 0–68-year-old offspring population and they show that the proportion of familial cases range from a low 0.4% and 0.5% for connective tissue and testicular cancer to the highest of 15.3% for prostate cancer. These figures depend on many parameters, including age of the population and prevalence of the disease. Familial proportions, as well as familial SIRs, are negatively affected by low penetrance of heritable cancer genes and polygenic effects, both of which would hide true familial risks and make clinical counseling less predictable.51, 52, 53
The familial risks for testicular cancer (4.26 by an affected parent and 9.28 by an affected brother) and Hodgkin's disease (4.88 by an affected parents) were highest in our study. Familial clustering of testicular cancer has been characterized earlier54, 55, 56, 57 but that of Hodgkin's disease has been not well known58, 59, 60, 61 and susceptibility genes have been identified for neither neoplasm. There was no large effect by age of onset, and clinical counseling should be indicated if any case was known in the immediate family. Other neoplasms that showed high familial risks from a parental or sibling proband, not discussed before, were esophageal, endocrine, pancreatic and renal cancer, and lymphohematopoietic neoplasms. When both a parent and a sibling was affected, the SIRs were high even for gastric and nervous system cancers. In the present Database, these familial triplets of an affected parent and 2 affected offspring are likely to signal high penetrant dominant effects.62 Almost all these familial cancers lack identified susceptibility genes.4 Even for sites, such as the stomach, pancreas and the kidney, for which heritable genes have been identified, they may not account for all the familial clustering noted.22, 63
The Family-Cancer Database was created by linking registers maintained at Statistics Sweden and the Swedish Cancer Registry.