- Email: [email protected]
Age-specific incidence of all neoplasms after colorectal cancer
Annals of Epidemiology 24 (2014) 785e788
Contents lists available at ScienceDirect
Annals of Epidemiology journal homepage: www.annalsofepidemiology.org
Brief communication
Age-specific incidence of all neoplasms after colorectal cancer Fabio Levi MD, MSc a, *, Lalao Randimbison ScD a, Rafael Blanc-Moya MD a, Carlo La Vecchia MD, MSc b, c a Department of Medicine and Community Health, Vaud Cancer Registry and Cancer Epidemiology Unit, Institute of Social and Preventive Medicine (IUMSP), Lausanne University Hospital, Lausanne, Switzerland b Department of Epidemiology, IRCCS-Istituto di Ricerche Farmacologiche “Mario Negri”, Milan, Italy c Department of Clinical Sciences and Community Health, University of Milan, Milan, Italy
a r t i c l e i n f o
a b s t r a c t
Article history: Received 4 February 2014 Accepted 22 July 2014 Available online 30 July 2014
Purpose: Patients diagnosed with a specific neoplasm tend to have a subsequent excess risk of the same neoplasm. The age incidence of a second neoplasm at the same site is approximately constant with age, and consequently the relative risk is greater at younger age. It is unclear whether such a line of reasoning can be extended from a specific neoplasm to the incidence of all neoplasms in subjects diagnosed with a defined neoplasm. Methods: We considered the age-specific incidence of all nonehormone-related epithelial neoplasms after a first primary colorectal cancer (n ¼ 9542) in the Vaud Cancer Registry data set. Results: In subjects with a previous colorectal cancer, the incidence rate of all other epithelial nonehormonerelated cancers was stable around 800 per 100,000 between age 30 and 60 years, and rose only about twofold to reach 1685 at age 70 to 79 years and 1826 per 100,000 at age 80 years or older. After excluding synchronous cancers, the rise was only about 1.5-fold, that is, from about 700 to 1000. In the general population, the incidence rate of all epithelial nonehormone-related cancers was 29 per 100,000 at age 30 to 39 years, and rose 30-fold to 883 per 100,000 at age 70 to 79 years. Excluding colorectal cancers, the rise of all nonehormone-related cancers was from 360 per 100,000 at age 40 to 49 years to 940 at age 70 to 79 years after colorectal cancer, and from 90 to 636 per 100,000 in the general population (i.e., 2.6- vs. 7.1-fold). Conclusions: The rise of incidence with age of all epithelial nonehormone-related second cancers after colorectal cancer is much smaller than in the general population. This can possibly be related to the occurrence of a single mutational event in a population of susceptible individuals, although alternative models are plausible within the complexity of the process of carcinogenesis. Ó 2014 Elsevier Inc. All rights reserved.
Keywords: Colorectal cancer Multiple malignancies Cancer registry Incidence
Introduction Patients with a specific neoplasm tend to have a subsequent excess risk of the same neoplasm. In terms of relative risk, such an excess risk is greater at younger age, whereas in absolute terms, the incidence of a second neoplasm at the same site is approximately constant with age [1]. Thus, women diagnosed with a breast cancer tend to have high and approximately constant subsequent breast cancer rates [2], and subjects diagnosed with head and neck [3,4], colorectal cancer [5,6], basal cell carcinoma, [7] or melanoma [8] of the skin tend to have high and constant subsequent incidence of the same neoplasm, respectively. These age distributions contrast with the well-known rise with The authors declare no conflict of interest. * Corresponding author. Department of Medicine and Community Health, Vaud Cancer Registry, Cancer Epidemiology Unit, Institute of Social and Preventive Medicine (IUMSP), Lausanne University Hospital, Route de la Corniche 10, 1010 Lausanne, Switzerland. Tel.: þ41 21 314 73 11; fax: þ41 21 314 73 17. E-mail address: [email protected] (F. Levi). 1047-2797/$ e see front matter Ó 2014 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.annepidem.2014.07.005
a power function of age of the incidence of nonehormone-related epithelial neoplasms in the general population [9]. Such a rise of cancer incidence with a power function of age has been interpreted, within the multistage theory of carcinogenesis, as an indicator that the process of carcinogenesis involves the accumulation of several stochastic somatic changes [10]. It is possible therefore that the constant incidence rate with age of second primary cancers is explained by the occurrence of a single mutational event in a population of susceptible individuals, within the simple multistage theory [11,12]. It is however unclear whether such a line of reasoning can be extended from the setting of second cancers at the same site as the primary tumor to the setting of second cancers at other anatomic sites. To test this hypothesis, we have considered the age-specific incidence of all epithelial nonehormone-related neoplasms after a first primary colorectal cancer in the Vaud Cancer Registry data set [6], and contrasted that age function with the age-specific rates of all epithelial nonehormone-related cancers in the same population.
786
F. Levi et al. / Annals of Epidemiology 24 (2014) 785e788
Materials and methods Data for the present study were derived from the Vaud Cancer Registry data set, which includes information concerning incident cases of malignant neoplasms occurring in the Swiss canton of Vaud (about 713,000 inhabitants, according to the December 2010 National Census) [6]. The Registry adheres to the rules of registration for the first and second primary cancers of the International Agency for Research on Cancer [13] and has been included in the International Agency for Research on Cancer’s Cancer Incidence in Five Continents volumes since 1982 [14]. Within colon, separate primary sites (i.e., second cancers) in the same patient were considered for left colon (International Classification of Diseases for Oncology, First Edition [ICD-O-1] topography code 153.1e153.3, 153.7) and right colon (ICD-O-1 153.0, 153.4e153.6). Within colorectum, in the same patient, up to three colorectal primaries could be registered, two for colon and one for rectum (ICD-O-1 154.0e154.1) [6]. Population-based incidence data have been available since 1974, and the present report includes data until 2010. The main information available comprises demographic characteristics of the patient, primary site, and histologic type of the tumor according to the standard ICD-O-1 (ICD-O-1 and both ICD-O-1 and ICD-O-3 since 2005) [15,16]. Passive and active follow-up were recorded, and each subsequent item of information concerning an already registered case was used to complete the record of that patient. Information from the death certificate was added to the registration file. After exclusion of basal and squamous cell carcinomas of the skin from the whole database, 9542 patients (5123 males and 4419 females) diagnosed with a first colorectal primary malignancy were abstracted from the Registry’s database and followed over the period 1974e2010 for the occurrence of a second epithelial nonehormonerelated neoplasms, emigration or death, contributing to a total of 45,900 person-years at risk. Nonmelanomatous skin cancers, germ cell, lymphoid and primary breast, gynecological and male genital neoplasms (i.e., sexehormone-related cancers) were excluded. Synchronous cancers were defined as those occurring within 2 months of diagnosis of the first cancer. Metachronous cancers were those occurring 2 months or later since diagnosis of the first one. We computed and contrasted the age-specific incidence rates of all second nonehormone-related epithelial cancers in subjects with a first epithelial colorectal cancer to those in the general population of the Canton of Vaud. We excluded basal and squamous cell carcinomas of the skin. We ran separate analyses including and excluding colorectal cancer. Results There were 591 second primary cancers after a first diagnosis of colorectal cancer. The most common sites of second neoplasms were colorectum (250, 42%), lung (75, 13%), bladder (44, 7%), cutaneous melanoma (36, 6%), and stomach (31, 5%). Of these, 177 were synchronous and 414 were metachronous. Table 1 and Figure 1A (on a logarithmic scale) give the agespecific incidence rates of all epithelial nonehormone-related cancer patients aged 30 years or older (n ¼ 588) after a diagnosis of first colorectal cancer, and in the general population. In subjects with a previous colorectal cancer, the incidence rate of all other cancers was stable around 800 per 100,000 between age 30 and 60 years and rose only about twofold to reach 1685 at age 70 to 79 years and 1826 per 100,000 at age 80 years or older. In the general population, the rate of epithelial nonehormone-related cancers was 29 per 100,000 at age 30 to 39 years and rose 30-fold to 883 per 100,000 at age 70 to 79 years. The difference in all cancer sites between subjects with colorectal cancer and the general population was somewhat greater in men than in women, but the patterns
Table 1 Incidence of second neoplasms* in patients diagnosed with colorectal cancer and in the general population. Vaud Cancer Registry, Switzerland, 1974e2010 Age (y)
Post colon* Rate/100,000
Males and females 30e39 804.7 40e49 866.2 50e59 808.8 60e69 1233.2 70e79 1684.7 80þ 1825.7 Males 30e49 967.0 50e59 901.3 60e69 1569.7 70e79 2445.9 80þ 1961.8 Females 30e49 719.6 50e59 696.5 60e69 838.5 70e79 1019.3 80þ 1735.9
General population* Rate ratio (no. of cases) [95% CIy]
Rate/100,000
Rate ratio
1z (6) 1.08 (28) [0.72e1.57] 1.01 (72) [0.79e1.28] 1.53 (163) [1.31e1.79] 2.07 (223) [1.82e2.38] 2.27 (96) [1.85e2.77]
29.2 108.6 326.4 628.6 882.7 852.8
1z 3.72 11.18 21.53 30.23 29.21
1z (21) 0.93 (44) [0.67e1.24] 1.62 (112) [1.32e1.96] 2.52 (151) [2.14e2.98] 2.03 (41) [1.45e2.76]
82.9 460.9 951.2 1365.2 1342.2
1z 5.56 11.47 16.47 16.19
55.1 198.8 353.5 545.0 617.9
1z 3.61 6.42 9.89 11.21
1z(13) 0.97 (28) 1.16 (51) 1.42 (72) 2.41 (55)
[0.65e1.41] [0.86e1.53] [1.11e1.80] [1.79e3.13]
* Second colorectal cancers included; basal and squamous cell carcinomas of the skin and nonepithelial and hormone-related neoplasms were excluded. y CI: confidence interval. z Reference category.
were similar. After exclusion of synchronous cancers, however, the rate ratios became similar in men and women. When the same analysis was conducted excluding (second) colorectal cancers (Table 2 and Fig. 1B), the pattern was similar in men, but the difference in rate ratios between subjects with a previous colorectal cancer and the general population was apparently smaller in women. Excluding colorectal cancers, the rise of all nonehormone-related cancers was from 360 per 100,000 at age 40 to 49 years to 940 at age 70 to 79 years after colorectal cancer, and from 90 to 636 per 100,000 in the general population (i.e., 2.6- vs. 7.1-fold). We ran sensitivity analyses excluding synchronous cancers (Table 3). After exclusion of synchronous cancers, the rise was about 1.5-fold, that is, from about 700 to about 1000 per 100,000. Overall, the results were similar after exclusion of second colorectal cancers (besides hormone-related and nonepithelial cancers), with a rise in all cancers combined from 360 per 100,000 at age 40 to 49 years to about 700 at age 70 years or older. As for the overall data set, the differences between men and women became smaller after exclusion of synchronous cancers. In an additional sensitivity analysis, we also split our data set in two periods, that is, 1974 to 1989 and 1990e2010, and again the results were consistent across calendar periods, with incidence of all other cancers after the diagnosis of first colorectal cancer rising only about twofold between age 40 to 49 years and 70 years or older in both periods (from 650 to 1250 per 100,000 in 1974e1989 and from 1000 to 2000 per 100,000 in 1990e2010). Discussion We observed that the incidence of several previous neoplasms of the same site is approximately constant with age for a large number of neoplasms [1], contrary to the well-known rise of the incidence of epithelial nonehormone-related cancers with the fourth to fifth power of age [9]. This translates in an approximately 30-fold difference in cancer incidence between age 30 and 80 years, as observed in the general population from major cancer registration systems worldwide (i.e., the SEER data set [17,18]) and in the present data set, too.
F. Levi et al. / Annals of Epidemiology 24 (2014) 785e788
787
Fig. 1. Age-specific incidence rates of all epithelial nonehormone-related cancers in subjects with colorectal cancer and in the general population (A) and after excluding second colorectal cancers (B). Vaud Cancer Registry, Switzerland, 1974e2010.
The interpretation of such a major difference in age-incidence patterns in subjects with a previous neoplasm of the same site compared with the general population has to be related to genetic susceptibility. Thus, the approximately constant incidence with age of a second cancer can possibly be explained by the occurrence of a single mutational event in a subpopulation of susceptible individuals [1,12,19]. An alternative model would predict that several, if not most, neoplasms in patients with colorectal cancer arose in a susceptible minority of subjects whose incidence had increased to a high constant level [12]. Given the complexity of the process of carcinogenesis [19], however, it is difficult to reconcile our observation to a single genetic or biologic mechanism. The present work extends the observation of a constant incidence with age for several neoplasms to the susceptibility not only of the same index neoplasm but also of a large number of epithelial cancers, particularly in the age range 30 to 60 years. We chose colorectal cancer as an index site because it is the most common not strongly hormone-related neoplasm in both sexes combined in this population and (as opposed to lung cancer) has an acceptable survival [20] thus providing an adequate number of
person-years at risk of developing a second neoplasm. We excluded not only nonepithelial (lymphoid and germ cell) neoplasms but also male and female hormone-related epithelial neoplasms, because these have a peculiar age distribution, that is, a leveling of the age curve after menopause for breast, ovarian, and cervical cancer and a sharp rise at later age for endometrial and prostatic cancer. This reflects the impact of sex hormones on these neoplasms, and the sex hormone level changes with age [21,22]. Colorectal cancer has been favorably related to female hormones [23e25], but its age curve in both sexes is similar to most nonesex hormoneerelated neoplasms. An intriguing observation is the apparent difference in rate ratios across age between men and women, which would indicate that at least part of such a general susceptibility to cancer can be sex related, that is, greater in men than in women. Exclusion of synchronous cancers, however, reduced such difference. This observation requires however independent confirmation. We excluded nonmelanomatous skin cancers [26] because their inclusion can be appreciably influenced by increased diagnostic attention after a colorectal cancer. Thus, our observation is
788
F. Levi et al. / Annals of Epidemiology 24 (2014) 785e788
Table 2 Incidence of second neoplasms in patients diagnosed with colorectal cancer and in the general population, after exclusion of second colorectal cancer. Vaud Cancer Registry, Switzerland, 1974e2010 Age (y)
Post colon* Rate/100,000
Males and females 30e39 253.1 40e49 359.9 50e59 499.7 60e69 788.9 70e79 941.2 80þ 895.2 Males 30e49 574.5 50e59 606.9 60e69 986.0 70e79 1574.1 80þ 991.8 Females 30e59 270.1 60e69 554.2 70e79 386.4 80þ 832.1
General population* Rate ratio (no. of cases)
Rate/100,000
Rate ratio
1y 1.42 1.97 3.12 3.72 3.54
(2) (12) (45) (106) (128) (48)
25.2 90.5 267.2 496.0 635.6 550.4
1y 3.59 10.60 19.68 25.22 21.84
1y 1.06 1.72 2.74 1.73
(13) (30) (72) (100) (21)
71.5 391.2 783.1 1094.4 942.8
1y 5.47 10.9 15.31 13.19
1y 2.05 1.43 3.08
(16) (34) (28) (27)
74.0 251.1 346.1 362.1
1y 3.39 4.68 4.89
* Second colorectal cancers excluded; basal and squamous cell carcinomas of the skin and nonepithelial and hormone-related neoplasms were excluded. y Reference category.
unlikely to be appreciably influenced by random error or bias because we had adequate number of second cancers after a diagnosis of colorectal cancer. Furthermore, our inference is based on simple computation of age-specific rates, which can hardly be affected by serious bias, particularly in a cancer registration system which has been in operation with recognized quality standards for over 30 years now [14,27]. The consistency of our findings after exclusion of synchronous cancers and across
Table 3 Incidence of second metachronous neoplasms in patients diagnosed with colorectal cancer. Vaud Cancer Registry, Switzerland, 1974e2010 Age (y)
Post colony
Post colon* Rate/100,000
Males and females 30e39 670.6 40e49 711.5 50e59 685.3 60e69 908.0 70e79 1110.6 80þ 1065.1 Males 30e49 921.0 50e59 778.4 60e69 1135.3 70e79 1717.3 80þ 1196.4 Females 30e59 532.0 60e69 641.3 70e79 580.5 80þ 978.6
Rate ratio (no. of cases)
Rate/100,000
Rate ratio (no. of cases)
1z 1.06 1.02 1.35 1.66 1.59
(5) (23) (61) (120) (147) (56)
253.1 359.9 466.4 647.5 698.6 708.7
1z 1.42 1.84 2.56 2.76 2.80
(2) (12) (42) (87) (95) (38)
1z 0.85 1.23 1.86 1.30
(20) (38) (81) (106) (25)
574.4 586.7 835.4 1196.5 755.7
1z 1.02 1.45 2.08 1.32
(13) (29) (61) (76) (16)
1z 1.21 1.09 1.84
(31) (35) (41) (31)
236.3 423.9 262.2 678.0
1z 1.79 1.11 2.87
(14) (26) (19) (22)
* Second colorectal cancers included; basal and squamous cell carcinomas of the skin and nonepithelial and hormone-related neoplasms were excluded. y Second colorectal cancers, basal and squamous cell carcinomas of the skin, and nonepithelial and hormone-related neoplasms were excluded. z Reference category.
separate strata of calendar periods also supports the validity of our observation. Acknowledgments The authors wish to thank Mrs. G. Garimoldi for editorial assistance. The work of CLV was supported by the Fondazione Italiana per la Ricerca sul Cancro (FIRC). References [1] Levi F, Boffetta P, La Vecchia C. High constant incidence rates of second primary neoplasms. Eur J Cancer Prev 2008;17:385e8. [2] Levi F, Randimbison L, Te VC, La Vecchia C. Contralateral breast cancer in Vaud, Switzerland. Int J Cancer 2001;93:612e3. [3] Levi F, Te VC, Randimbison L, Maspoli M, La Vecchia C. Second primary oral and pharyngeal cancers in subjects diagnosed with oral and pharyngeal cancer. Int J Cancer 2006;119:2702e4. [4] Bosetti C, Scelo G, Chuang SC, Tonita JM, Tamaro S, Jonasson JG, et al. High constant incidence rates of second primary cancers of the head and neck: a pooled analysis of 13 cancer registries. Int J Cancer 2011;129:173e9. [5] Levi F, Randimbison L, Te VC, La Vecchia C. Re: Effect of age on risk of second primary colorectal cancer. J Natl Cancer Inst 2002;94:529. author reply 29e30. [6] Levi F, Randimbison L, Blanc-Moya R, Maspoli-Conconi M, Rosato V, Bosetti C, et al. High constant incidence of second primary colorectal cancer. Int J Cancer 2013;132:1679e82. [7] Levi F, Randimbison L, Maspoli M, Te VC, La Vecchia C. High incidence of second basal cell skin cancers. Int J Cancer 2006;119:1505e7. [8] Levi F, Randimbison L, Te VC, La Vecchia C. High constant incidence rates of second cutaneous melanomas. Int J Cancer 2005;117:877e9. [9] Armitage P, Doll R. The age distribution of cancer and a multi-stage theory of carcinogenesis. Br J Cancer 2004;91:1983e9. [10] Armitage P, Doll R. A two-stage theory of carcinogenesis in relation to the age distribution of human cancer. Br J Cancer 1957;11:161e9. [11] Easton D. Breast cancerenot just whether but when? Nat Genet 2000;26:390e1. [12] Peto J, Mack TM. High constant incidence in twins and other relatives of women with breast cancer. Nat Genet 2000;26:411e4. [13] IARC. International Agency for Research on Cancer. International rules for multiple primary cancers (ICD-O third edition). Eur J Cancer Prev 2005;14:307e8. [14] Levi F, Delachaux A, Junod B, Wietlisbach V. Cancer incidence in Vaud, Switzerland (1975-1977). In: Waterhouse J, Muir C, Shanmugaratnam K, Powell J, editors. Cancer Incidence in Five Continents, vol. IV. Lyon: IARC; 1982. p. 546e9. IARC Scientific Publications No 42. [15] WHO. World Health Organization. International Classification of Diseases for Oncology, ICD-O. Geneva: World Health Organization; 1976. p. 131. [16] WHO. World Health Organization. International Classification of Diseases for Oncology. 3rd ed. Geneva: World Health Organization; 2000. p. 240. [17] Matanoski G, Tao XG, Almon L, Adade AA, Davies-Cole JD. Demographics and tumor characteristics of colorectal cancers in the United States, 1998-2001. Cancer 2006;107(5 Suppl):1112e20. [18] Rim SH, Seef L, Ahmed F, King JB, Coughlin SS. Colorectal cancer incidence in the United States, 1999-2004. Cancer 2009;115:1967e76. [19] Hanahan D, Weinberg RA. Hallmarks of cancer: the next generation. Cell 2011;144:646e74. [20] Levi F, Randimbison L, Te VC, Franceschi S, La Vecchia C. Trends in survival for patients diagnosed with cancer in Vaud, Switzerland, between 1974 and 1993. Ann Oncol 2000;11:957e63. [21] Doll R. Epidemiology of cancer: current perspectives. Am J Epidemiol 1976;104:396e404. [22] Doll R. A epidemiological perspective of the biology of cancer. Cancer Res 1978;38:3573e83. [23] Campbell-Thompson M, Lynch IJ, Bhardwaj B. Expression of estrogen (ER) subtypes and ERbeta isoforms in colon cancer. Cancer Res 2001;61:632e40. [24] Newcomb PA, Zheng Y, Chia VM, Morimoto LM, Doria-Rose VP, Templeton A, et al. Estrogen plus progestin use, microsatellite instability, and the risk of colorectal cancer in women. Cancer Res 2007;67:7534e9. [25] Bosetti C, Bravi F, Negri E, La Vecchia C. Oral contraceptives and colorectal cancer risk: a systematic review and meta-analysis. Hum Reprod Update 2009;15:489e98. [26] Levi F, Franceschi S, Te VC, Randimbison L, La Vecchia C. Trends of skin cancer in the Canton of Vaud, 1976-92. Br J Cancer 1995;72:1047e53. [27] Levi F, Te VC, Blanc-Moya R, Randimbison L, Choffat R. Cancer incidence in Vaud (2003-2007). In: Forman D, Bray F, Brewster DH, Gombe Mbalawa C, Kohler P, Pineros M, et al., editors. Cancer Incidence in Five Continents, Vol. X. Lyon: IARC; 2013. electronic version, http://ci5.iarc.fr. last accessed November 2013.
Contents lists available at ScienceDirect
Annals of Epidemiology journal homepage: www.annalsofepidemiology.org
Brief communication
Age-specific incidence of all neoplasms after colorectal cancer Fabio Levi MD, MSc a, *, Lalao Randimbison ScD a, Rafael Blanc-Moya MD a, Carlo La Vecchia MD, MSc b, c a Department of Medicine and Community Health, Vaud Cancer Registry and Cancer Epidemiology Unit, Institute of Social and Preventive Medicine (IUMSP), Lausanne University Hospital, Lausanne, Switzerland b Department of Epidemiology, IRCCS-Istituto di Ricerche Farmacologiche “Mario Negri”, Milan, Italy c Department of Clinical Sciences and Community Health, University of Milan, Milan, Italy
a r t i c l e i n f o
a b s t r a c t
Article history: Received 4 February 2014 Accepted 22 July 2014 Available online 30 July 2014
Purpose: Patients diagnosed with a specific neoplasm tend to have a subsequent excess risk of the same neoplasm. The age incidence of a second neoplasm at the same site is approximately constant with age, and consequently the relative risk is greater at younger age. It is unclear whether such a line of reasoning can be extended from a specific neoplasm to the incidence of all neoplasms in subjects diagnosed with a defined neoplasm. Methods: We considered the age-specific incidence of all nonehormone-related epithelial neoplasms after a first primary colorectal cancer (n ¼ 9542) in the Vaud Cancer Registry data set. Results: In subjects with a previous colorectal cancer, the incidence rate of all other epithelial nonehormonerelated cancers was stable around 800 per 100,000 between age 30 and 60 years, and rose only about twofold to reach 1685 at age 70 to 79 years and 1826 per 100,000 at age 80 years or older. After excluding synchronous cancers, the rise was only about 1.5-fold, that is, from about 700 to 1000. In the general population, the incidence rate of all epithelial nonehormone-related cancers was 29 per 100,000 at age 30 to 39 years, and rose 30-fold to 883 per 100,000 at age 70 to 79 years. Excluding colorectal cancers, the rise of all nonehormone-related cancers was from 360 per 100,000 at age 40 to 49 years to 940 at age 70 to 79 years after colorectal cancer, and from 90 to 636 per 100,000 in the general population (i.e., 2.6- vs. 7.1-fold). Conclusions: The rise of incidence with age of all epithelial nonehormone-related second cancers after colorectal cancer is much smaller than in the general population. This can possibly be related to the occurrence of a single mutational event in a population of susceptible individuals, although alternative models are plausible within the complexity of the process of carcinogenesis. Ó 2014 Elsevier Inc. All rights reserved.
Keywords: Colorectal cancer Multiple malignancies Cancer registry Incidence
Introduction Patients with a specific neoplasm tend to have a subsequent excess risk of the same neoplasm. In terms of relative risk, such an excess risk is greater at younger age, whereas in absolute terms, the incidence of a second neoplasm at the same site is approximately constant with age [1]. Thus, women diagnosed with a breast cancer tend to have high and approximately constant subsequent breast cancer rates [2], and subjects diagnosed with head and neck [3,4], colorectal cancer [5,6], basal cell carcinoma, [7] or melanoma [8] of the skin tend to have high and constant subsequent incidence of the same neoplasm, respectively. These age distributions contrast with the well-known rise with The authors declare no conflict of interest. * Corresponding author. Department of Medicine and Community Health, Vaud Cancer Registry, Cancer Epidemiology Unit, Institute of Social and Preventive Medicine (IUMSP), Lausanne University Hospital, Route de la Corniche 10, 1010 Lausanne, Switzerland. Tel.: þ41 21 314 73 11; fax: þ41 21 314 73 17. E-mail address: [email protected] (F. Levi). 1047-2797/$ e see front matter Ó 2014 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.annepidem.2014.07.005
a power function of age of the incidence of nonehormone-related epithelial neoplasms in the general population [9]. Such a rise of cancer incidence with a power function of age has been interpreted, within the multistage theory of carcinogenesis, as an indicator that the process of carcinogenesis involves the accumulation of several stochastic somatic changes [10]. It is possible therefore that the constant incidence rate with age of second primary cancers is explained by the occurrence of a single mutational event in a population of susceptible individuals, within the simple multistage theory [11,12]. It is however unclear whether such a line of reasoning can be extended from the setting of second cancers at the same site as the primary tumor to the setting of second cancers at other anatomic sites. To test this hypothesis, we have considered the age-specific incidence of all epithelial nonehormone-related neoplasms after a first primary colorectal cancer in the Vaud Cancer Registry data set [6], and contrasted that age function with the age-specific rates of all epithelial nonehormone-related cancers in the same population.
786
F. Levi et al. / Annals of Epidemiology 24 (2014) 785e788
Materials and methods Data for the present study were derived from the Vaud Cancer Registry data set, which includes information concerning incident cases of malignant neoplasms occurring in the Swiss canton of Vaud (about 713,000 inhabitants, according to the December 2010 National Census) [6]. The Registry adheres to the rules of registration for the first and second primary cancers of the International Agency for Research on Cancer [13] and has been included in the International Agency for Research on Cancer’s Cancer Incidence in Five Continents volumes since 1982 [14]. Within colon, separate primary sites (i.e., second cancers) in the same patient were considered for left colon (International Classification of Diseases for Oncology, First Edition [ICD-O-1] topography code 153.1e153.3, 153.7) and right colon (ICD-O-1 153.0, 153.4e153.6). Within colorectum, in the same patient, up to three colorectal primaries could be registered, two for colon and one for rectum (ICD-O-1 154.0e154.1) [6]. Population-based incidence data have been available since 1974, and the present report includes data until 2010. The main information available comprises demographic characteristics of the patient, primary site, and histologic type of the tumor according to the standard ICD-O-1 (ICD-O-1 and both ICD-O-1 and ICD-O-3 since 2005) [15,16]. Passive and active follow-up were recorded, and each subsequent item of information concerning an already registered case was used to complete the record of that patient. Information from the death certificate was added to the registration file. After exclusion of basal and squamous cell carcinomas of the skin from the whole database, 9542 patients (5123 males and 4419 females) diagnosed with a first colorectal primary malignancy were abstracted from the Registry’s database and followed over the period 1974e2010 for the occurrence of a second epithelial nonehormonerelated neoplasms, emigration or death, contributing to a total of 45,900 person-years at risk. Nonmelanomatous skin cancers, germ cell, lymphoid and primary breast, gynecological and male genital neoplasms (i.e., sexehormone-related cancers) were excluded. Synchronous cancers were defined as those occurring within 2 months of diagnosis of the first cancer. Metachronous cancers were those occurring 2 months or later since diagnosis of the first one. We computed and contrasted the age-specific incidence rates of all second nonehormone-related epithelial cancers in subjects with a first epithelial colorectal cancer to those in the general population of the Canton of Vaud. We excluded basal and squamous cell carcinomas of the skin. We ran separate analyses including and excluding colorectal cancer. Results There were 591 second primary cancers after a first diagnosis of colorectal cancer. The most common sites of second neoplasms were colorectum (250, 42%), lung (75, 13%), bladder (44, 7%), cutaneous melanoma (36, 6%), and stomach (31, 5%). Of these, 177 were synchronous and 414 were metachronous. Table 1 and Figure 1A (on a logarithmic scale) give the agespecific incidence rates of all epithelial nonehormone-related cancer patients aged 30 years or older (n ¼ 588) after a diagnosis of first colorectal cancer, and in the general population. In subjects with a previous colorectal cancer, the incidence rate of all other cancers was stable around 800 per 100,000 between age 30 and 60 years and rose only about twofold to reach 1685 at age 70 to 79 years and 1826 per 100,000 at age 80 years or older. In the general population, the rate of epithelial nonehormone-related cancers was 29 per 100,000 at age 30 to 39 years and rose 30-fold to 883 per 100,000 at age 70 to 79 years. The difference in all cancer sites between subjects with colorectal cancer and the general population was somewhat greater in men than in women, but the patterns
Table 1 Incidence of second neoplasms* in patients diagnosed with colorectal cancer and in the general population. Vaud Cancer Registry, Switzerland, 1974e2010 Age (y)
Post colon* Rate/100,000
Males and females 30e39 804.7 40e49 866.2 50e59 808.8 60e69 1233.2 70e79 1684.7 80þ 1825.7 Males 30e49 967.0 50e59 901.3 60e69 1569.7 70e79 2445.9 80þ 1961.8 Females 30e49 719.6 50e59 696.5 60e69 838.5 70e79 1019.3 80þ 1735.9
General population* Rate ratio (no. of cases) [95% CIy]
Rate/100,000
Rate ratio
1z (6) 1.08 (28) [0.72e1.57] 1.01 (72) [0.79e1.28] 1.53 (163) [1.31e1.79] 2.07 (223) [1.82e2.38] 2.27 (96) [1.85e2.77]
29.2 108.6 326.4 628.6 882.7 852.8
1z 3.72 11.18 21.53 30.23 29.21
1z (21) 0.93 (44) [0.67e1.24] 1.62 (112) [1.32e1.96] 2.52 (151) [2.14e2.98] 2.03 (41) [1.45e2.76]
82.9 460.9 951.2 1365.2 1342.2
1z 5.56 11.47 16.47 16.19
55.1 198.8 353.5 545.0 617.9
1z 3.61 6.42 9.89 11.21
1z(13) 0.97 (28) 1.16 (51) 1.42 (72) 2.41 (55)
[0.65e1.41] [0.86e1.53] [1.11e1.80] [1.79e3.13]
* Second colorectal cancers included; basal and squamous cell carcinomas of the skin and nonepithelial and hormone-related neoplasms were excluded. y CI: confidence interval. z Reference category.
were similar. After exclusion of synchronous cancers, however, the rate ratios became similar in men and women. When the same analysis was conducted excluding (second) colorectal cancers (Table 2 and Fig. 1B), the pattern was similar in men, but the difference in rate ratios between subjects with a previous colorectal cancer and the general population was apparently smaller in women. Excluding colorectal cancers, the rise of all nonehormone-related cancers was from 360 per 100,000 at age 40 to 49 years to 940 at age 70 to 79 years after colorectal cancer, and from 90 to 636 per 100,000 in the general population (i.e., 2.6- vs. 7.1-fold). We ran sensitivity analyses excluding synchronous cancers (Table 3). After exclusion of synchronous cancers, the rise was about 1.5-fold, that is, from about 700 to about 1000 per 100,000. Overall, the results were similar after exclusion of second colorectal cancers (besides hormone-related and nonepithelial cancers), with a rise in all cancers combined from 360 per 100,000 at age 40 to 49 years to about 700 at age 70 years or older. As for the overall data set, the differences between men and women became smaller after exclusion of synchronous cancers. In an additional sensitivity analysis, we also split our data set in two periods, that is, 1974 to 1989 and 1990e2010, and again the results were consistent across calendar periods, with incidence of all other cancers after the diagnosis of first colorectal cancer rising only about twofold between age 40 to 49 years and 70 years or older in both periods (from 650 to 1250 per 100,000 in 1974e1989 and from 1000 to 2000 per 100,000 in 1990e2010). Discussion We observed that the incidence of several previous neoplasms of the same site is approximately constant with age for a large number of neoplasms [1], contrary to the well-known rise of the incidence of epithelial nonehormone-related cancers with the fourth to fifth power of age [9]. This translates in an approximately 30-fold difference in cancer incidence between age 30 and 80 years, as observed in the general population from major cancer registration systems worldwide (i.e., the SEER data set [17,18]) and in the present data set, too.
F. Levi et al. / Annals of Epidemiology 24 (2014) 785e788
787
Fig. 1. Age-specific incidence rates of all epithelial nonehormone-related cancers in subjects with colorectal cancer and in the general population (A) and after excluding second colorectal cancers (B). Vaud Cancer Registry, Switzerland, 1974e2010.
The interpretation of such a major difference in age-incidence patterns in subjects with a previous neoplasm of the same site compared with the general population has to be related to genetic susceptibility. Thus, the approximately constant incidence with age of a second cancer can possibly be explained by the occurrence of a single mutational event in a subpopulation of susceptible individuals [1,12,19]. An alternative model would predict that several, if not most, neoplasms in patients with colorectal cancer arose in a susceptible minority of subjects whose incidence had increased to a high constant level [12]. Given the complexity of the process of carcinogenesis [19], however, it is difficult to reconcile our observation to a single genetic or biologic mechanism. The present work extends the observation of a constant incidence with age for several neoplasms to the susceptibility not only of the same index neoplasm but also of a large number of epithelial cancers, particularly in the age range 30 to 60 years. We chose colorectal cancer as an index site because it is the most common not strongly hormone-related neoplasm in both sexes combined in this population and (as opposed to lung cancer) has an acceptable survival [20] thus providing an adequate number of
person-years at risk of developing a second neoplasm. We excluded not only nonepithelial (lymphoid and germ cell) neoplasms but also male and female hormone-related epithelial neoplasms, because these have a peculiar age distribution, that is, a leveling of the age curve after menopause for breast, ovarian, and cervical cancer and a sharp rise at later age for endometrial and prostatic cancer. This reflects the impact of sex hormones on these neoplasms, and the sex hormone level changes with age [21,22]. Colorectal cancer has been favorably related to female hormones [23e25], but its age curve in both sexes is similar to most nonesex hormoneerelated neoplasms. An intriguing observation is the apparent difference in rate ratios across age between men and women, which would indicate that at least part of such a general susceptibility to cancer can be sex related, that is, greater in men than in women. Exclusion of synchronous cancers, however, reduced such difference. This observation requires however independent confirmation. We excluded nonmelanomatous skin cancers [26] because their inclusion can be appreciably influenced by increased diagnostic attention after a colorectal cancer. Thus, our observation is
788
F. Levi et al. / Annals of Epidemiology 24 (2014) 785e788
Table 2 Incidence of second neoplasms in patients diagnosed with colorectal cancer and in the general population, after exclusion of second colorectal cancer. Vaud Cancer Registry, Switzerland, 1974e2010 Age (y)
Post colon* Rate/100,000
Males and females 30e39 253.1 40e49 359.9 50e59 499.7 60e69 788.9 70e79 941.2 80þ 895.2 Males 30e49 574.5 50e59 606.9 60e69 986.0 70e79 1574.1 80þ 991.8 Females 30e59 270.1 60e69 554.2 70e79 386.4 80þ 832.1
General population* Rate ratio (no. of cases)
Rate/100,000
Rate ratio
1y 1.42 1.97 3.12 3.72 3.54
(2) (12) (45) (106) (128) (48)
25.2 90.5 267.2 496.0 635.6 550.4
1y 3.59 10.60 19.68 25.22 21.84
1y 1.06 1.72 2.74 1.73
(13) (30) (72) (100) (21)
71.5 391.2 783.1 1094.4 942.8
1y 5.47 10.9 15.31 13.19
1y 2.05 1.43 3.08
(16) (34) (28) (27)
74.0 251.1 346.1 362.1
1y 3.39 4.68 4.89
* Second colorectal cancers excluded; basal and squamous cell carcinomas of the skin and nonepithelial and hormone-related neoplasms were excluded. y Reference category.
unlikely to be appreciably influenced by random error or bias because we had adequate number of second cancers after a diagnosis of colorectal cancer. Furthermore, our inference is based on simple computation of age-specific rates, which can hardly be affected by serious bias, particularly in a cancer registration system which has been in operation with recognized quality standards for over 30 years now [14,27]. The consistency of our findings after exclusion of synchronous cancers and across
Table 3 Incidence of second metachronous neoplasms in patients diagnosed with colorectal cancer. Vaud Cancer Registry, Switzerland, 1974e2010 Age (y)
Post colony
Post colon* Rate/100,000
Males and females 30e39 670.6 40e49 711.5 50e59 685.3 60e69 908.0 70e79 1110.6 80þ 1065.1 Males 30e49 921.0 50e59 778.4 60e69 1135.3 70e79 1717.3 80þ 1196.4 Females 30e59 532.0 60e69 641.3 70e79 580.5 80þ 978.6
Rate ratio (no. of cases)
Rate/100,000
Rate ratio (no. of cases)
1z 1.06 1.02 1.35 1.66 1.59
(5) (23) (61) (120) (147) (56)
253.1 359.9 466.4 647.5 698.6 708.7
1z 1.42 1.84 2.56 2.76 2.80
(2) (12) (42) (87) (95) (38)
1z 0.85 1.23 1.86 1.30
(20) (38) (81) (106) (25)
574.4 586.7 835.4 1196.5 755.7
1z 1.02 1.45 2.08 1.32
(13) (29) (61) (76) (16)
1z 1.21 1.09 1.84
(31) (35) (41) (31)
236.3 423.9 262.2 678.0
1z 1.79 1.11 2.87
(14) (26) (19) (22)
* Second colorectal cancers included; basal and squamous cell carcinomas of the skin and nonepithelial and hormone-related neoplasms were excluded. y Second colorectal cancers, basal and squamous cell carcinomas of the skin, and nonepithelial and hormone-related neoplasms were excluded. z Reference category.
separate strata of calendar periods also supports the validity of our observation. Acknowledgments The authors wish to thank Mrs. G. Garimoldi for editorial assistance. The work of CLV was supported by the Fondazione Italiana per la Ricerca sul Cancro (FIRC). References [1] Levi F, Boffetta P, La Vecchia C. High constant incidence rates of second primary neoplasms. Eur J Cancer Prev 2008;17:385e8. [2] Levi F, Randimbison L, Te VC, La Vecchia C. Contralateral breast cancer in Vaud, Switzerland. Int J Cancer 2001;93:612e3. [3] Levi F, Te VC, Randimbison L, Maspoli M, La Vecchia C. Second primary oral and pharyngeal cancers in subjects diagnosed with oral and pharyngeal cancer. Int J Cancer 2006;119:2702e4. [4] Bosetti C, Scelo G, Chuang SC, Tonita JM, Tamaro S, Jonasson JG, et al. High constant incidence rates of second primary cancers of the head and neck: a pooled analysis of 13 cancer registries. Int J Cancer 2011;129:173e9. [5] Levi F, Randimbison L, Te VC, La Vecchia C. Re: Effect of age on risk of second primary colorectal cancer. J Natl Cancer Inst 2002;94:529. author reply 29e30. [6] Levi F, Randimbison L, Blanc-Moya R, Maspoli-Conconi M, Rosato V, Bosetti C, et al. High constant incidence of second primary colorectal cancer. Int J Cancer 2013;132:1679e82. [7] Levi F, Randimbison L, Maspoli M, Te VC, La Vecchia C. High incidence of second basal cell skin cancers. Int J Cancer 2006;119:1505e7. [8] Levi F, Randimbison L, Te VC, La Vecchia C. High constant incidence rates of second cutaneous melanomas. Int J Cancer 2005;117:877e9. [9] Armitage P, Doll R. The age distribution of cancer and a multi-stage theory of carcinogenesis. Br J Cancer 2004;91:1983e9. [10] Armitage P, Doll R. A two-stage theory of carcinogenesis in relation to the age distribution of human cancer. Br J Cancer 1957;11:161e9. [11] Easton D. Breast cancerenot just whether but when? Nat Genet 2000;26:390e1. [12] Peto J, Mack TM. High constant incidence in twins and other relatives of women with breast cancer. Nat Genet 2000;26:411e4. [13] IARC. International Agency for Research on Cancer. International rules for multiple primary cancers (ICD-O third edition). Eur J Cancer Prev 2005;14:307e8. [14] Levi F, Delachaux A, Junod B, Wietlisbach V. Cancer incidence in Vaud, Switzerland (1975-1977). In: Waterhouse J, Muir C, Shanmugaratnam K, Powell J, editors. Cancer Incidence in Five Continents, vol. IV. Lyon: IARC; 1982. p. 546e9. IARC Scientific Publications No 42. [15] WHO. World Health Organization. International Classification of Diseases for Oncology, ICD-O. Geneva: World Health Organization; 1976. p. 131. [16] WHO. World Health Organization. International Classification of Diseases for Oncology. 3rd ed. Geneva: World Health Organization; 2000. p. 240. [17] Matanoski G, Tao XG, Almon L, Adade AA, Davies-Cole JD. Demographics and tumor characteristics of colorectal cancers in the United States, 1998-2001. Cancer 2006;107(5 Suppl):1112e20. [18] Rim SH, Seef L, Ahmed F, King JB, Coughlin SS. Colorectal cancer incidence in the United States, 1999-2004. Cancer 2009;115:1967e76. [19] Hanahan D, Weinberg RA. Hallmarks of cancer: the next generation. Cell 2011;144:646e74. [20] Levi F, Randimbison L, Te VC, Franceschi S, La Vecchia C. Trends in survival for patients diagnosed with cancer in Vaud, Switzerland, between 1974 and 1993. Ann Oncol 2000;11:957e63. [21] Doll R. Epidemiology of cancer: current perspectives. Am J Epidemiol 1976;104:396e404. [22] Doll R. A epidemiological perspective of the biology of cancer. Cancer Res 1978;38:3573e83. [23] Campbell-Thompson M, Lynch IJ, Bhardwaj B. Expression of estrogen (ER) subtypes and ERbeta isoforms in colon cancer. Cancer Res 2001;61:632e40. [24] Newcomb PA, Zheng Y, Chia VM, Morimoto LM, Doria-Rose VP, Templeton A, et al. Estrogen plus progestin use, microsatellite instability, and the risk of colorectal cancer in women. Cancer Res 2007;67:7534e9. [25] Bosetti C, Bravi F, Negri E, La Vecchia C. Oral contraceptives and colorectal cancer risk: a systematic review and meta-analysis. Hum Reprod Update 2009;15:489e98. [26] Levi F, Franceschi S, Te VC, Randimbison L, La Vecchia C. Trends of skin cancer in the Canton of Vaud, 1976-92. Br J Cancer 1995;72:1047e53. [27] Levi F, Te VC, Blanc-Moya R, Randimbison L, Choffat R. Cancer incidence in Vaud (2003-2007). In: Forman D, Bray F, Brewster DH, Gombe Mbalawa C, Kohler P, Pineros M, et al., editors. Cancer Incidence in Five Continents, Vol. X. Lyon: IARC; 2013. electronic version, http://ci5.iarc.fr. last accessed November 2013.