Specil Report Risk of Colorectl Cncer by Subsite in Swedish Prostte Cncer Cohort Yunxi Lu, MD, PhD, Rickrd Ljung, MD, PhD, Ann Mrtling, MD, PhD, nd Mts Lindbld, MD, PhD Bckground: The reltionship between sex hormone relted tretment for prostte cncer nd the risk of colorectl cncer is controversil. Methods: A prostte cncer cohort ws initited from the Swedish Cncer Registry of ptients dignosed between 1961 nd 2008. Ptients dignosed with prostte cncer between 1961 nd 1980 were generlly treted with estrogen. The cohort dignosed between 1981 nd 2008 ws further divided into 3 subcohorts of orchiectomy, prosttectomy, nd other tretment. Stndrdized incidence rtios (SIRs) for developing colorectl denocrcinom were estimted nd 95% confidence intervls (CIs) were used to compre reltive risk mong these ptients nd the generl mle popultion. Results: Of 601,542 person-yers of follow-up, 1,698 cses of colorectl denocrcinom were identified. Compred with the generl mle popultion, no ssocition ws detected in the cohort dignosed between 1961 nd 1980, wheres n incresed risk of colorectl denocrcinom ws observed mong ptients dignosed with prostte cncer who received tretments other thn estrogen. Following bilterl orchiectomy, the SIR ws 1.30 (95% CI: 1.14 1.47); fter prosttectomy, the SIR ws 1.22 (95% CI: 1.04 1.43); mong those who received tretment other thn estrogen, the SIR ws 1.37 (95% CI: 1.29 1.45). The incresed risks were more pprent in cses of denocrcinom of the distl colon nd rectum thn in the proximl colon. Conclusions: Ptients with prostte cncer undergoing bilterl orchiectomy, prosttectomy, or other tretments, including ntindrogen therpy nd rdition, my be t incresed risk for colorectl denocrcinom. Introduction The role of the influence of sex hormones in the etiology of colorectl cncer (CRC) deserves ttention, prticulrly regrding potentil differences between the loctions of such cncer. Bsed on seprte embryologicl origins, nd thus potentilly divergent cusl pthwys, the colorectl intestine is often divided into the proximl colon (cecum, scending colon, nd trnsverse colon), distl colon (descending colon nd sigmoid colon), nd rectum. Accumulting evidence from epidemiologicl studies hs demonstrted the differences in risk-fctor profiles between men nd women of cncers in the colorectl subsites with regrd to dietry fctors, 1-4 obesity, 5 nd physicl From the Deprtment of Moleculr Medicine nd Surgery (YL, RL, AM) nd the Unit of Epidemiology t the Institute of Environmentl Medicine (RL), Krolinsk Institutet, Stockholm, Sweden, nd the Deprtment of Epidemiology nd Biosttistics (YL, ML), Imperil College, London, Englnd. Address correspondence to Yunxi Lu, MD, PhD, Deprtment of Moleculr Medicine nd Surgery, Krolinsk Institutet, SE-171 76, Stockholm, Sweden. E-mil: yunxi.lu@ki.se Submitted July 3, 2014; ccepted September 26, 2014. The uthors received finncil support from the Swedish Reserch Council nd the Swedish Society of Medicine for this study. ctivity. 6,7 The incidence of cncer of the proximl colon is 10% to 20% higher in women thn in men t ll ges, wheres men hve higher incidence of cncer in the distl colon nd rectum. 8,9 It is possible tht sex hormones could differently influence cncers of the proximl colon, distl colon, nd rectum. Epidemiologicl studies in women hve shown tht higher levels of the femle sex hormones estrogen nd progesterone re ssocited with lower risk of developing CRC, prticulrly when exogenous steroid hormones re used, 10,11 but few studies hve ssocited sex hormones with risk of colorectl denocrcinom by subsite in men. 12,13 Previously, hormone therpy ws mjor tretment pproch for prostte cncer, lthough hormone therpy does produce dverse events. 14-16 Estrogen therpy nd ndrogen deprivtion therpy (ADT; including orchiectomy nd gondotropin-relesing hormone [GnRH] gonists) re 2 types of sex hormone therpies used in ptients with prostte cncer. Theoreticlly, ptients with prostte cncer treted with sex hormones my constitute n idel nturl humn model for evluting the ssocition between externl sex hormones nd the risk of CRC. In 2010, study found tht ptients with prostte cncer receiving ADT April 2015, Vol. 22, No. 2 Cncer Control 263
hd n incresed risk of CRC compred with ptients not receiving ADT. 17 Due to confounding by indiction, this ssocition is still controversil. Further studies compring ptients with prostte cncer nd the generl popultion my provide more evidence. Methods Becuse of the nture of Swedish ntionl helth registries (see below for detiled description of ech registry used), we were ble to identify ll of the ptients dignosed with prostte cncer in Sweden between 1961 nd 2008 nd followed them up to the development of second denocrcinom in the colon or rectum. The ptients with prostte cncer were grouped into 4 tretment cohorts. The im of this study ws to determine whether prostte cncer tretments, prticulrly sex hormone therpy, were ssocited with risk for CRC. The study ws pproved by the Regionl Ethicl Review Bord in Stockholm, Sweden. Dt Sources This study ws bsed on dt from 3 ntionwide Swedish popultion-bsed registries, ie, the cncer, ptient, nd totl popultion registries. The personl identity number, unique 10-digit identifiction number ssigned to ech resident of Sweden, ws used to link ll of the ptients between the registries. Cncer Registry: The cncer registry ws initited in 1958 nd includes the dte of dignosis, tumor site, nd histologicl type of ll mlignnt tumors dignosed in Sweden. All newly dignosed tumors in the country must be reported to the cncer registry by clinicin nd pthologist or cytologist. The completeness of the registry pproches 98% to 100%, 18 nd 99% of ll tumors re morphologiclly verified. 19 The cohorts included ll men registered with first dignosis of prostte cncer. The yers 1958 to 1960 were excluded to void inclusion of the prevlence of cncer cses when the registry begn. All ptients with ny dignosis of other previous cncer were excluded. Any subsequent cncer fter prostte cncer ws identified throughout the cncer registry, including the outcome of colorectl denocrcinom (proximl colon, distl colon, unspecified colon, nd rectum). Ptient Registry: The ptient registry ws initited in 1964 nd covered 2 Swedish counties; strting in 1987, the registry covered 100% of Sweden. 20 This registry hs chieved ccurcy nd surgicl completeness rtes of 95% nd 98%, respectively. 21 Informtion on the surgicl procedures following the dignosis of prostte cncer ws collected from this registry. The Swedish Clssifiction of Opertions nd Mjor Procedures hs been included in the ptient registry since 1964, nd it ws used to identify ptients undergoing bilterl orchiectomy nd prosttectomy. Totl Popultion Registry: The totl popultion registry provides complete nd continuously updted informtion on dtes of emigrtion, immigrtion, births, nd deths since 1961. Dt of censored cohort members who died or emigrted during follow-up were collected from this registry. However, these ptients were censored from the dte of first emigrtion or exct dte of deth. Study Cohorts The methods for treting prostte cncer chnged in Sweden during the 1980s, with estrogen therpy being the predominnt tretment strtegy prior to 1980. 22 Bilterl orchiectomy ws lso n vilble stndrd tretment for prostte cncer between the yers 1950 nd 1980 in mny countries, but orchiectomy ws not populr in Sweden during these yers. 23 Other tretments, such s ADT (orchiectomy nd GnRH nlogues), rdicl prosttectomy, nd rdicl rdiotherpy, incresed fter the 1980s. 22,24-26 The number of ptients who did not receive tretment (ctive surveillnce nd wtchful witing) ws low. 27 Therefore, the cohort ws divided into mutully exclusive subcohorts bsed on the definition of sex hormone tretments following prostte cncer dignosis (Fig). The cohort dignosed prior to 1981 included persons dignosed with prostte cncer before 1981; second cohort of ptients included those dignosed with prostte cncer between Jnury 1, 1981, nd December 31, 2008. The lst subcohort ws further divided into ptients who underwent bilterl orchiectomy, ptients who underwent prosttectomy, nd ptients treted minly by GnRH therpy, rdiotherpy, or both ssigned to the other tretment group. Follow-Up The follow-up of the cohort members strted from the dte prostte cncer ws newly dignosed or the strt of specific tretments in the cohort, nd ptients were followed-up until ny of the following end points: dignosis of ny cncer (except nonmelnom skin cncer), emigrtion, ge 85 yers, deth, or end of the study period, whichever cme first. The end of the study period for the cohort before the 1980s ws set to December 31, 1980, nd to December 31, 2008, for ll the other cohorts. 22 The cutoff dte (December 31, 1980) for the end of the first study period or the beginning of the second study period ws rbitrrily selected bsed on the previously reported chnges of tretment in Sweden. Sttisticl Anlyses Person-yers were clculted from the dte of prostte cncer dignosis (or specific tretment, eg, orchiectomy) to the dte of the study s end point. 264 Cncer Control April 2015, Vol. 22, No. 2
Dignosis of Prostte Cncer (N = 149,743) Estrogen-Exposed Cohort Dignosis of prostte cncer (1961 1980) (n = 33,373) Nonexposed Estrogen Cohort Dignosis of prostte cncer (1981 2008) (n = 116,370) Bilterl Orchiectomy (n = 21,917) Prosttectomy (n = 16,521) Other Tretments (n = 77,932) Fig. Flow chrt for the prostte cncer cohorts (1961 2008). Stndrdized incidence rtios (SIRs) were estimted s mesure of reltive risk when compring the dt from those dignosed with prostte cncer nd the corresponding generl mle popultion. The rtio ws clculted bsed on the observed number divided by the expected number of newly dignosed cses of CRC. The expected number of cses ws clculted by multiplying the observed number of person-yers by the ge- nd clendr yer specific incidence rtes with 5-yer intervls of the entire Swedish mle popultion. SIRs nd 95% confidence intervls (CIs) were clculted with the ssumption tht the number of cses followed Poisson distribution. SIRs were seprtely estimted for cncers of the proximl colon, distl colon, unspecified colon, nd rectum in ech of the cohorts. We lso performed sensitivity nlyses with respect to the seprte cohorts, which we ssumed were exposed to vrious prostte cncer tretments. For the cohort dignosed between 1961 nd 1980, we nlyzed the cohort members who were followed from 1961 to 1975, with the end point being 5 yers erlier thn the prior ssumed cutoff dte. For the 3 cohorts dignosed fter 1980, we performed nlyses on ptients who entered into the cohort from 1986 through 2008, with the strt point being 5 yers lter thn the prior ssumed cutoff dte. In the min nlysis, we excluded the first yer of follow-up fter the dignosis of prostte cncer becuse cncers dignosed in the first yer were prticulrly prone to detection bis nd unlikely to be relted to the tretment of prostte cncer. All tests were 2-sided with significnce level of 0.05. Anlyses were performed using the SAS sttisticl pckge, version 9.0 (SAS Institute, Cry, North Crolin). Results A totl of 149,743 ptients with prostte cncer were included in the finl study cohort nd informtion for the specific cohorts is shown in Tble 1. The cohort dignosed between 1961 nd 1980 ws composed of 33,373 ptients with prostte cncer who were followed-up for 111,809 person-yers until 1980, nd the totl cohort dignosed with prostte cncer between 1981 nd 2008 included 116,370 ptients with prostte cncer nd 489,733 person-yers of follow-up since 1981. During follow-up, we identified totl of 1,698 cses of colorectl denocrcinom. Among them, 487 cses of denocrcinom were locted in the proximl colon, 453 cses in the distl colon, 132 cses in n unspecified site of the colon, nd 626 cses in the rectum. Age t entry nd ge t dignosis of colorectl denocrcinom were significntly different mong the groups (ll P vlues <.01). Those in the prosttectomy cohort were younger thn ptients in the other cohorts. Estrogen Exposure Compred with the generl popultion, the cohort dignosed between 1961 nd 1980 ws not ssocited with the overll development of colorectl denocrcinom (SIR 0.98; 95% CI: 0.85 1.12) or when nlyzed by subsite (Tble 2). Results suggested sttisticlly nonsignificnt decresed risk of denocrcinom of the proximl colon, but these results were minly observed mong ptients with short follow-up; however, nonsignificnt trend of incresed risk with yers of follow-up (P vlue for trend =.09) in the proximl colon ws identified. Orchiectomy Between 1981 nd 2008, totl of 21,917 individuls underwent bilterl orchiectomy nd hd n verge follow-up period of 3.4 yers (Tbles 1 nd 3). An overll incresed risk of colorectl denocrcinom ws detected in ptients with 1 to 9 yers of follow-up, but this risk decrese fter 10 yers of follow-up. A significntly incresed risk April 2015, Vol. 22, No. 2 Cncer Control 265
Tble 1. Bsic Chrcteristics of the Prostte Cncer Tretment Groups Estrogen Exposed (1961 1980) Specific Tretment Cohorts Nonexposed Estrogen (1981 2008) Orchiectomy Prosttectomy Other Tretment No. of cohort members 33,373 21,917 16,521 77,932 Totl person-y of follow-up 111,809 75,043 75,700 338,990 Averge ge t entry, y 71.9 ± 7.3 74.9 ± 6.4 63.8 ± 5.9 71.1 ± 7.3 Averge follow-up, y 3.4 3.4 4.6 4.3 No. of colorectl denocrcinom cses 198 247 156 1097 No. of colorectl denocrcinom cses t first yer of follow-up 74 97 16 332 Averge ge t dignosis of colorectl denocrcinom, y 76.2 ± 6.2 77.8 ± 4.7 70.5 ± 6.1 76.4 ± 5.6 Person-y incidence (95% confidence intervl ) b Colorectl denocrcinom 177 (152.4 201.8) 329 (288.1 370.2) 206 (173.7 238.4) 323 (304.5 342.8) Proximl colon denocrcinom 46 (33.1 58.1) 77 (57.4 97.2) 40 (25.4 53.8) 102 (91.9 113.4) Distl colon denocrcinom 53 (39.3 66.2) 85 (64.4 106.2) 63 (45.5 81.3) 83 (73.5 92.9) Rectl colon 59 (44.8 73.2) 124 (98.7 149.1) 94 (72.0 115.6) 117 (105.3 128.3) Unspecified colon denocrcinom 20 (11.5 27.9) 43 (27.9 57.4) 9 (2.4 16.1) 21 (16.1 25.8) Clcultion bsed on the first yer of follow-up hd been excluded. b Incidence counted s per 100,000 person-y. Tble 2. SIRs nd 95% CIs of Colorectl in the Estrogen-Exposed Cohort Compred With the Generl Popultion (1961 1980) Colorectl 198 202 0.98 (0.85 1.12) Age t dignosis of prostte cncer, y 41 64 25 29 0.86 (0.55 1.26) 65 74 104 105 0.99 (0.81 1.20) 75 69 68 1.02 (0.79 1.28) of the Proximl Colon 51 58 0.88 (0.66 1.16) 3 8 0.39 (0.08 1.13) 29 30 0.98 (0.66 1.41) 19 20 0.93 (0.56 1.45) of the Distl Colon 59 56 1.05 (0.80 1.35) 8 8 1.00 (0.43 1.98) 34 30 1.15 (0.80 1.61) 17 19 0.90 (0.52 1.44) of the Rectum 66 74 0.89 (0.69 1.13) 11 12 0.94 (0.47 1.69) 29 39 0.74 (0.50 1.06) 26 24 1.10 (0.72 1.61) Unspecified Colorectl 22 14 1.60 (1.00 2.42) 3 2 1.69 (0.35 4.93) 12 7 1.77 (0.88 2.98) 7 5 1.41 (0.57 2.90) P vlue for trend 0.52.31.63.043.72 Follow-up, y 1 4 129 135 0.96 (0.80 1.14) 5 9 59 54 1.10 (0.83 1.41) 10 10 14 0.74 (0.35 1.36) 29 39 0.74 (0.50 1.06) 17 15 1.14 (0.66 1.82) 5 4 1.37 (0.44 3.19) 40 39 1.04 (0.74 1.41) 16 14 1.12 (0.64 1.81) 3 3 0.88 (0.18 2.58) 44 48 0.92 (0.67 1.24) 20 21 0.95 (0.58 1.46) 2 6 0.35 (0.04 1.27) 16 10 1.67 (0.95 2.71) 6 3 1.76 (0.64 3.83) 0 0 0 P vlue for trend.99.09.98.34.49 First yer of follow-up ws excluded. CI = confidence intervl, Exp = expected, Obs = observed, SIRS = stndrdized incidence rtio. ws observed for the overll rte of colorectl denocrcinom (SIR 1.30; 95% CI: 1.14 1.47) nd for ech subsite except for the proximl colon. Prosttectomy The prosttectomy cohort ws composed of 16,521 ptients who underwent prosttectomy between 1981 nd 2008 nd hd n verge follow-up period of 4.6 yers (Tbles 1 nd 4). Similrly incresed risks of colorectl denocrcinom overll nd by specific subsites were observed mong ptients in the prosttectomy cohort. 266 Cncer Control April 2015, Vol. 22, No. 2
Tble 3. SIRs nd 95% CIs of Colorectl in the Prostte Cncer Cohort With Orchiectomy Compred With the Generl Popultion (1981 2008) Colorectl 247 190 1.30 (1.14 1.47) Age t dignosis of prostte cncer, y 41 64 26 17 1.55 (1.01 2.27) 65 74 132 97 1.36 (1.14 1.61) 75 89 76 1.17 (0.94 1.44) of the Proximl Colon 58 55 1.06 (0.80 1.36) 2 4 0.46 (0.06 1.66) 34 27 1.24 (0.86 1.74) 22 23 0.95 (0.59 1.43) of the Distl Colon 64 44 1.44 (1.11 1.84) 10 4 2.48 (1.19 4.57) 32 23 1.40 (0.96 1.98) 22 17 1.26 (0.79 1.91) of the Rectum 93 74 1.26 (1.01 1.54) 13 7 1.85 (0.98 3.16) 44 38 1.14 (0.83 1.53) 36 28 1.26 (0.89 1.75) Unspecified Colorectl 32 17 1.91 (1.31 2.70) 1 1 0.73 (0.02 4.06) 22 9 2.57 (1.61 3.89) 9 7 1.32 (0.60 2.50) P vlue for trend.15.99.14.51.49 Follow-up, y 1 4 158 127 1.24 (1.06 1.45) 5 9 75 49 1.53 (1.20 1.91) 10 14 14 1.00 (0.55 1.68) 38 36 1.05 (0.74 1.44) 15 14 1.05 (0.59 1.72) 5 4 1.17 (0.38 2.73) 37 29 1.26 (0.89 1.73) 24 11 2.10 (1.34 3.12) 3 3 0.88 (0.18 2.57) 58 50 1.16 (0.88 1.50) 30 19 1.59 (1.07 2.27) 5 5 0.96 (0.31 2.24) 25 11 2.22 (1.44 3.28) 6 4 1.36 (0.50 2.95) 1 1 0.92 (0.02 5.11) P vlue for trend.77.86.47.59.18 First yer of follow-up ws excluded. CI = confidence intervl, Exp = expected, Obs = observed, SIRS = stndrdized incidence rtio. Specificlly, n incresed SIR ws found for the distl colon (1.44; 95% CI: 1.06 1.91) nd rectum (1.36; 95% CI: 1.06 1.71). Other Tretments Among ptients dignosed with prostte cncer fter 1980, totl of 77,932 did undergo bilterl orchiectomy or prosttectomy during the study period (see Tble 1). Ptients in the other tretment cohort hd n overll incresed risk of colorectl denocrcinom (SIR 1.37; 95% CI: 1.29 1.45) nd for ll subsites (Tble 5). Specificlly, n incresed risk of denocrcinom in the proximl colon ws ssocited with ge (P vlue for trend <.01); decresed risk ws observed in ssocition with yers of follow-up (P vlue for trend =.02; see Tble 5). Sensitivity Anlyses Anlyses of the cohort dignosed between 1961 nd 1975 did not identify n ssocition between estrogen tretment nd overll or subsite risk of CRC, finding consistent with the results from the originlly defined cohort (dignosed between 1961 nd 1980). In the orchiectomy cohort, which included ptients dignosed between 1986 nd 2008, incresed SIRs of CRC were found, in line with the results shown in Tble 3. In the prosttectomy nd other tretment cohorts, similrly incresed nd stronger SIRs were observed when compred with the generl mle popultion. Discussion This ntionwide Swedish cohort study identified n incresed risk of colorectl denocrcinom for ll ntomicl loctions mong ptients exposed to bilterl orchiectomy, ADT, rdicl prosttectomy, nd rdiotherpy. No cler ssocition ws detected mong individuls exposed to estrogen. Tretment for prostte cncer dignosed fter 1980 my be ssocited with n incresed risk of colorectl denocrcinom, implying possible connection to ADT, one of the most common tretments in Sweden used fter 1980. 27 The incresed risk of colorectl denocrcinom found mong ptients undergoing bilterl orchiectomy could be relted to n effect of surgicl ndrogen deprivtion, wheres the incresed risk seen in ptients not undergoing orchiectomy or prosttectomy could be due to mediclly induced ndrogen deprivtion (ie, GnRH therpy). Higher SIRs were observed in the sensitivity nlyses for the prosttectomy nd other tretment cohorts, prticulrly in those with short-term follow-up. The results lso indicted tht other tretments such s GnRH therpy my ply role in the risk of CRC, lthough the probbility vlue for such trend did not rech sttisticl significnce. Alterntively, crcinogenic effect of rdiotherpy for prostte cncer my be present; however, this represents exposure for which we hve no dt. Consistent results hve lso been found in US study April 2015, Vol. 22, No. 2 Cncer Control 267
Tble 4. SIRs nd 95% CIs of Colorectl in the Prosttectomy Cohort Compred With the Generl Popultion (1981 2008) Colorectl 156 127 1.22 (1.04 1.43) Age t dignosis of prostte cncer, y 41 64 73 63 1.16 (0.91 1.46) 65 74 79 62 1.27 (1.00 1.58) 75 4 2 1.89 (0.51 4.83) of the Proximl Colon 30 35 0.86 (0.58 1.23) 13 16 0.79 (0.42 1.36) 17 18 0.95 (0.55 1.52) 0 1 0.00 (0.00 5.75) of the Distl Colon 48 33 1.44 (1.06 1.91) 19 16 1.17 (0.70 1.82) 27 16 1.63 (1.07 2.37) 2 1 4.00 (0.48 14.43) of the Rectum 71 52 1.36 (1.06 1.71) 36 27 1.33 (0.93 1.84) 33 24 1.35 (0.93 1.89) 2 1 2.50 (0.30 9.04) Unspecified Colorectl 7 7 1.04 (0.42 2.15) 5 3 1.59 (0.52 3.71) 2 3 0.59 (0.07 2.13) 0 0 0.00 (0.00 20.85) P vlue for trend.41.86.12.72.19 Follow-up, y 1 4 89 68 1.31 (1.05 1.61) 5 9 49 40 1.22 (0.90 1.61) 10 18 19 0.95 (0.56 1.50) 16 18 0.89 (0.51 1.44) 11 11 0.98 (0.49 1.76) 3 6 0.53 (0.11 1.54) 28 18 1.59 (1.05 2.29) 14 11 1.32 (0.72 2.21) 6 5 1.18 (0.43 2.57) 41 29 1.41 (1.02 1.92) 22 16 1.35 (0.85 2.05) 8 7 1.12 (0.48 2.20) 4 4 1.13 (0.31 2.90) 2 2 0.92 (0.11 3.34) 1 1 0.98 (0.02 5.48) P vlue for trend.24.56.44.57.84 First yer of follow-up ws excluded. CI = confidence intervl, Exp = expected, Obs = observed, SIRS = stndrdized incidence rtio. A Tble 5. SIRs nd 95% CIs of Colorectl in Other Tretment Cohorts Compred With the Generl Popultion (1981 2008) Colorectl 1,097 800 1.37 (1.29 1.45) Age t dignosis of prostte cncer, y 41 64 133 108 1.23 (1.03 1.45) 65 74 585 429 1.36 (1.26 1.48) 75 379 263 1.44 (1.30 1.59) of the Proximl Colon 348 234 1.49 (1.33 1.65) 28 31 1.10 (0.75 1.56) 169 123 1.37 (1.17 1.60) 148 83 1.79 (1.51 2.10) tht found strong ssocition between orchiectomy nd colorectl denocrcinom. 17 The elevted of the Distl Colon 282 203 1.39 (1.23 1.56) 34 28 1.24 (0.86 1.73) 158 111 1.43 (1.21 1.67) 90 65 1.38 (1.11 1.70) of the Rectum 396 309 1.28 (1.16 1.41) 61 46 1.32 (1.01 1.70) 219 167 1.31 (1.15 1.50) 116 96 1.21 (1.00 1.45) Unspecified Colorectl 71 54 1.31 (1.03 1.66) 7 7 1.07 (0.43 2.21) 39 28 1.38 (0.98 1.88) 25 19 1.30 (0.84 1.92) P vlue for trend.12.00.72.48.81 Follow-up b, y 1 4 694 489 1.42 (1.32 1.53) 5 9 321 239 1.34 (1.20 1.50) 10 82 72 1.14 (0.91 1.42) 227 141 1.61 (1.41 1.84) 99 71 1.39 (1.13 1.69) 22 22 1.00 (0.62 1.51) 175 123 1.42 (1.22 1.64) 81 61 1.32 (1.05 1.64) 26 19 1.40 (0.92 2.06) 244 191 1.28 (1.12 1.45) 123 91 1.35 (1.12 1.61) 29 27 1.09 (0.73 1.56) 48 34 1.42 (1.05 1.88) 18 16 1.14 (0.68 1.81) 5 5 1.11 (0.36 2.58) P vlue for trend.07.02.76.75.40 Study cohorts excluded ptients who underwent orchiectomy nd prosttectomy. b First yer of follow-up ws excluded. CI = confidence intervl, Exp = expected, Obs = observed, SIRS = stndrdized incidence rtio. risk (increse in CRC incidence of 30% 40% in men exposed to ADT) ws noticeble even fter djust- 268 Cncer Control April 2015, Vol. 22, No. 2
ing for potentil confounders such s dibetes, obesity, nd rdiotherpy. 17 Although tht study hd robust dt regrding orchiectomy nd GnRH therpy, it included bis from confounding by indiction nd its prticipnts were older. 17 Our study, which mesured reltive risk bsed on comprison between the study cohort groups nd the generl popultion, further demonstrted the findings of the US study. 17 Estrogen nd Androgen We did not find ny ssocition between estrogen tretment nd colorectl denocrcinom. Previous reports, which were bsed on observtionl dt nd results from clinicl trils, hve suggested tht women exposed to exogenous estrogen, including hormone replcement therpy or orl contrceptives, hve decresed risk of CRC. 12,27,28 However, the estrogen hypothesis hs seldom been evluted in men. The reltionship between estrogen nd ndrogen is intriguing. Estrogens re produced with ndrogens s precursors. Progesterone is the first importnt sexul steroid formed in the body. The ndrogens (dehydroepindrosterone, ndrostenedione, nd testosterone) rise therefter, wheres the estrogens (estrone nd estrdiol) pper only during the finl stge. Estrogen nd ndrogen my ply similr role in crcinogenesis mong women nd men, respectively. Estrogen nd ndrogen were postulted to prevent tumor growth by preventing insulin nd insulin-like growth fctor from binding to their receptors. 29 By contrst, n incresed risk of colorectl denocrcinom due to ndrogen deprivtion could be possible s the results from niml studies suggest tht ndrogen my hve protective effect on the development of CRC. 30,31 Androgen receptors hve lso been found more frequently in the norml s opposed to cncerous mucos in the colon. 31,32 Thus, orchiectomy or GnRH tretment would directly decrese ndrogen. Specific Subsite Our results suggest differences in risk my exist between the proximl nd distl colon. The decresing trend of SIRs ssocited with denocrcinom in the proximl colon over follow-up yers could indicte differing ssocition t tht subsite. Select genetic or physiologicl mechnisms might explin this difference in risk ptterns of denocrcinom in the proximl nd distl colon. The proximl colon origintes from the midgut, but the distl colon nd rectum originte from the hindgut. Reserch suggests tht different genetic pthwys to CRC dominte the proximl nd the distl segments of the bowel. 33 These genetic-dependent pthwys re influenced by different sex-relted fctors. 34 Estrogen receptors re distributed differently in the proximl nd distl colon. 35 Furthermore, chromosoml instbility nd microstellite instbility 2 forms of genetic instbility in CRC mostly occur in the proximl nd distl colon, respectively, wrrnting the genetic bsis. 36 Limittions Strengths of this study include its ntionwide nd popultion-bsed cohort design identified from ntionl registries, its lrge prostte cncer cohort sizes, nd its long nd complete follow-up times; however, informtion on individul prostte cncer tretment ws limited. Dt on bilterl orchiectomy, which is common ADT, ws vilble in the ptient registry from 1964 onwrd, but we hd no detiled informtion on other hormonl tretments such s estrogen or ntindrogen medictions. We specified the cohorts bsed on time periods nd specific tretments. The results of the sensitivity nlyses showed fir consistency with the reported results bsed on the selected cutoff dtes. Another concern is the introduction of prostte-specific ntigen (PSA), which could chnge the profiles of ptients with prostte cncer. 28 PSA screening ws introduced in Sweden between 1995 nd 2000 nd stedily incresed from 1% to 10%. 37 Less ggressive prostte cncer might hve been dignosed fter the introduction of PSA testing. However, this would hve little influence on our results becuse we strtified the nlysis in ll 3 subcohorts. Tretments for the first 2 cohorts were defined, wheres influences on the third cohort could hve been overlooked. Considering the lower proportion of PSA screening nd the lrge size of this cohort, PSA screening should not engender substntil influence on the totl results. It is possible tht CRC screening might influence the dignosis of CRC in ptients with prostte cncer. In Sweden, CRC screening commenced in the 1980s but ptients were limited; the officilly orgnized screening of CRC begn in Stockholm in 2008 nd in the rest of Sweden in 2014, 38,39 so CRC screening would not hve substntilly ffected the results of our study. Furthermore, we did not find studies to indicte tht colonoscopy ws incresed in the prostte cncer survivor group in Sweden. It my be possible tht ptients receiving rdiotherpy for prostte cncer my lso hve hd bowel symptoms. Becuse SIRs were clculted bsed on comprison of the prostte cncer cohort with the generl popultion t the sme period, such n influence would be minor. The dverse events of hormone therpy re concerning nd my contribute to n incresed risk of CRC, 17 but such informtion cnnot be specified in the current study. Further djustment of obesity nd dibetes did not chnge the ssocition of hormone therpy nd CRC in one study. 17 Lstly, no informtion ws provided on wht my April 2015, Vol. 22, No. 2 Cncer Control 269
be considered to be confounding fctors, including fmilil bckground nd dietry hbits. However, it is unlikely tht these fctors could explin our results, s it is likely they re eqully distributed between ptients regrdless of prostte cncer tretment. Although the rndomized clinicl tril is the gold stndrd for this type of reserch, its use ws not fesible for our study due to ethicl issues. Conclusions Results from this popultion-bsed cohort study suggest tht ndrogen deprivtion therpy, including bilterl orchiectomy, my increse the risk of colorectl denocrcinom in men, lthough confounders of rdiotherpy or prostte cncer could not be excluded. Further studies re wrrnted to elucidte this potentil ssocition. References 1. Lrsson SC, Rfter J, Holmberg L, et l. Red met consumption nd risk of cncers of the proximl colon, distl colon nd rectum: the Swedish Mmmogrphy Cohort. Int J Cncer. 2005;113(5):829-834. 2. Koushik A, Hunter DJ, Spiegelmn D, et l. Fruits, vegetbles, nd colon cncer risk in pooled nlysis of 14 cohort studies. J Ntl Cncer Inst. 2007;99(19):1471-1483. 3. Kim MK, Sski S, Otni T, et l; Jpn Public Helth Center-bsed Prospective Study Group. Dietry ptterns nd subsequent colorectl cncer risk by subsite: prospective cohort study. Int J Cncer. 2005;115(5):790-798. 4. Cho A, Thun MJ, Connell CJ, et l. Met consumption nd risk of colorectl cncer. JAMA. 2005;293(2):172-182. 5. Lke I, Thune I, Selmer R, et l. A prospective study of body mss index, weight chnge, nd risk of cncer in the proximl nd distl colon. Cncer Epidemiol Biomrkers Prev. 2010;19(6):1511-1522. 6. Boyle T, Fritschi L, Heyworth J, et l. Long-term sedentry work nd the risk of subsite-specific colorectl cncer. Am J Epidemiol. 2011;173(10):1183-1191. 7. Mordi T, Gridley G, Björk J, et l. Occuptionl physicl ctivity nd risk for cncer of the colon nd rectum in Sweden mong men nd women by ntomic subsite. Eur J Cncer Prev. 2008;17(3):201-208. 8. McMichel AJ, Potter JD. Host fctors in crcinogenesis: certin bile-cid metbolic profiles tht selectively increse the risk of proximl colon cncer. J Ntl Cncer Inst. 1985;75(2):185-191. 9. Potter JD, Slttery ML, Bostick RM, et l. Colon cncer: review of the epidemiology. Epidemiol Rev. 1993;15(2):499-545. 10. Chlebowski RT, Wctwski-Wende J, Ritenbugh C, et l. Estrogen plus progestin nd colorectl cncer in postmenopusl women. N Engl J Med. 2004;350(10):991-1004. 11. Berl V, Bnks E, Reeves G. Evidence from rndomised trils on the long-term effects of hormone replcement therpy. Lncet. 2002;360(9337):942-944. 12. Troisi R, Schirer C, Chow WH, et l. Reproductive fctors, orl contrceptive use, nd risk of colorectl cncer. Epidemiology. 1997;8(1):75-79. 13. Delellis Henderson K, Dun L, Sullivn-Hlley J, et l. Menopusl hormone therpy use nd risk of invsive colon cncer: the Cliforni Techers Study. Am J Epidemiol. 2010;171(14):415-425. 14. Seible DM, Gu X, Hytt AS, et l. Weight gin on ndrogen deprivtion therpy: which ptients re t highest risk? Urology. 2014;83(6):1316-1321. 15. Jmes H III, Aleksic I, Bienz MN, et l. Comprison of frcture risk ssessment tool score to bone minerl density for estimting frcture risk in ptients with dvnced prostte cncer on ndrogen deprivtion therpy. Urology. 2014;84(1):164-168. 16. Schmidt-Hnsen M, Hoskin P, Kirkbride P, et l. Hormone nd rdiotherpy versus hormone or rdiotherpy lone for non-metsttic prostte cncer: systemtic review with met-nlyses. Clin Oncol (R Coll Rdiol). 2014;26(10):e21-e46. 17. Gillessen S, Templeton A, Mrr G, et l. Risk of colorectl cncer in men on long-term ndrogen deprivtion therpy for prostte cncer. J Ntl Cncer Inst. 2010;102(23):1760-1770. 18. Brlow L, Westergren K, Holmberg L, et l. The completeness of the Swedish Cncer Register: smple survey for yer 1998. Act Oncologic. 2009;48(1):27-33. 19. Sttistics Sweden. Officil sttistics of Sweden. http://www.scb.se/en_/ About-us/Officil-Sttistics-of-Sweden. Accessed Jnury 7, 2015. 20. Nessén T, Prker R, Persson I, et l. Time trends in incidence rtes of first hip frcture in the Uppsl Helth Cre Region, Sweden, 1965-1983. Am J Epidemiol. 1989;130(2):289-299. 21. Nilsson AC, Spetz CL, Crsjo K, et l. Relibility of the hospitl registry. The dignostic dt re better thn their reputtion [in Swedish]. Lkrtidningen. 1994;91(7):598, 603-605. 22. Giertz G. Urology in Sweden: 1940-1990 [in Swedish]. Sydsven Medicinhist Sllsk Arssk. 1996;(suppl 21):1-284. 23. Lindbld M, Ye W, Rubio C, et l. Estrogen nd risk of gstric cncer: protective effect in ntionwide cohort study of ptients with prostte cncer in Sweden. Cncer Epidemiol Biomrkers Prev. 2004;13(12):2203-2207. 24. McLeod DG. Hormonl therpy: historicl perspective to future directions. Urology. 2003;61(2 suppl 1):3-7. 25. Adolfsson J, Grmo H, Vrenhorst E, et l. Clinicl chrcteristics nd primry tretment of prostte cncer in Sweden between 1996 nd 2005. Scnd J Urol Nephrol. 2007;41(6):456-477. 26. Cox RL, Crwford ED. Estrogens in the tretment of prostte cncer. J Urol. 1995;154(6):1991-1998. 27. Onkologiskt Centrum Norr Regionen. The Ntionl Prostte Cncer Register (NPCR) in Sweden 2004 2008. Uppsl, Sweden: Onkologiskt Centrum; 2010. http://www.cncercentrum.se/globl/dignoser/prosttcncer/rpporter/prosttc_rpport_04_08.pdf. Accessed Jnury 7, 2015. 28. Lilj H, Ulmert D, Vickers AJ. Prostte-specific ntigen nd prostte cncer: prediction, detection nd monitoring. Nt Rev Cncer. 2008;8:268-278. 29. Slttery ML, Bllrd-Brbsh R, Edwrds S, et l. Body mss index nd colon cncer: n evlution of the modifying effects of estrogen (United Sttes). Cncer Cuses Control. 2003;14(1):75-84. 30. Chn AT, Giovnnucci EL. Primry prevention of colorectl cncer. Gstroenterology. 2010;138(6):2029-2043e10. 31. Izbicki JR, Hmilton SR, Wmbch G, et l. Effects of ndrogen mnipultions on chemiclly induced colonic tumours nd on mcroscopiclly norml colonic mucos in mle Sprgue-Dwley rts. Br J Cncer. 1990;61(2):235-240. 32. Ro CV, Tokumo K, Rigotty J, et l. Chemoprevention of colon crcinogenesis by dietry dministrtion of piroxicm, lph-difluoromethylornithine, 16 lph-fluoro-5-ndrosten-17-one, nd ellgic cid individully nd in combintion. Cncer Res. 1991;51(17):4528-4534. 33. Sugi T, Hbno W, Jio YF, et l. Anlysis of moleculr ltertions in left- nd right-sided colorectl crcinoms revels distinct pthwys of crcinogenesis: proposl for new moleculr profile of colorectl crcinoms. J Molec Dig. 2006;8(2):193-201. 34. Breivik J, Lothe RA, Meling GI, et l. Different genetic pthwys to proximl nd distl colorectl cncer influenced by sex-relted fctors. Int J Cncer. 1997;74(6):664-669. 35. Jssm N, Bell SM, Speirs V, et l. Loss of expression of oestrogen receptor bet in colon cncer nd its ssocition with Dukes stging. Oncol Rep. 2005;14(1):17-21. 36. Lindblom A. Different mechnisms in the tumorigenesis of proximl nd distl colon cncers. Curr Opin Oncol. 2001;13(1):63-69. 37. Vickers AJ, Ulmert D, Sjoberg DD, et l. Strtegy for detection of prostte cncer bsed on reltion between prostte specific ntigen t ge 40-55 nd long term risk of metstsis: cse-control study. BMJ. 2013;346:f2023. 38. Blom J, Kilpeläinen S, Hultcrntz R, et l. Five-yer experience of orgnized colorectl cncer screening in Swedish popultion - incresed complince with ge, femle gender, nd subsequent screening round. J Med Screen. 2014;21(3):144-150. 39. Hkm M, Hoff G, Kronborg O, et l. Screening for colorectl cncer. Act Oncol. 2005;44(5):425-439. 270 Cncer Control April 2015, Vol. 22, No. 2