original article

Annals of Oncology 18: 1408–1413, 2007 doi:10.1093/annonc/mdm127

Prognosis for long-term survivors of cancer M. L. G. Janssen-Heijnen1*, S. Houterman1,2, V. E. P. P. Lemmens1,3, H. Brenner4, E. W. Steyerberg3 & J. W. W. Coebergh1,3 1 Eindhoven Cancer Registry, Comprehensive Cancer Centre South, AE Eindhoven; 2MMC Academy, Ma´xima Medical Centre, Veldhoven; 3Department of Public Health, Erasmus MC—University Medical Centre Rotterdam, Rotterdam, The Netherlands; 4Division of Clinical Epidemiology and Aging Research, German Cancer Research Centre, Heidelberg, Germany

original article

Received 1 November 2006; revised 14 March 2007; accepted 14 March 2007

Background: Many cancer patients who have already survived some time want to know about their prognosis, given the pre-condition that they are still alive. We described and interpreted population-based conditional 5-year relative survival rates. Patients and methods: The long-standing Eindhoven Cancer Registry collects data on all patients diagnosed with cancer in the southern part of the Netherlands. Patients aged 25–74 years, diagnosed between 1960 and 2004, were included. Conditional 5-year relative survival was computed for every additional year survived (follow-up period 1980–2004). Results: For patients with colorectal cancer, cutaneous melanoma or stage I breast cancer, conditional 5-year relative survival was >95% after having survived 3–15 years. However, for stomach, lung, stage II or III breast, prostate cancer or Hodgkin lymphoma, conditional 5-year relative survival did not exceed 75–94%. Initial differences in survival at diagnosis between age, gender and stage groups largely disappeared after having survived for 5–10 years. Conclusion: Prognosis for patients with cancer generally improved with each year survived. Patients with colorectal cancer, cutaneous melanoma or stage I breast cancer hardly exhibit any excess mortality after 3–15 years, whereas for patients with other tumours survival remained poorer than for the general population. Insight into conditional survival is especially useful for (ex)patients, who may use this information to plan their remaining life. Key words: cancer, conditional survival, long-term prognosis, period survival, population-based

introduction Patients who have survived for a certain amount of time after cancer diagnosis often want to know about their prognosis at that time. Standard survival curves at diagnosis of cancer are too negative, because these are also based on patients who die within the first years. Conditional survival analysis is a method for estimating the survival rate, given the precondition of having already survived a certain length of time [1]. For example, 5-year relative survival at diagnosis of stomach cancer is 20%. After 5 years, 80% of the patients have died from stomach cancer. Patients who are still alive at 5 years belong to a group of survivors with at that moment a much better prognosis than at diagnosis. Conditional survival rates can provide patients and treating physicians with more relevant information for personal health-related future planning [2]. For estimating conditional survival rates [1, 3-8], a long-term and complete follow-up is needed. Some previous studies have presented conditional survival rates, most of them with data from the Surveillance, Epidemiology and End Results (SEER) in *Correspondence to: M. L. G. Janssen-Heijnen, Eindhoven Cancer Registry, PO Box 231, 5600 AE Eindhoven, The Netherlands. Tel: +31 40 2971616; Fax: +31 40 2971610; E-mail: [email protected]

ª 2007 European Society for Medical Oncology

the United States. Only one of them [8], pertaining to patients diagnosed in the 1970s, has presented conditional survival rates for those still alive >5 years after diagnosis and one study has estimated the conditional probability of cure up to 15 years after diagnosis of cutaneous melanoma [9]. In this article we determined conditional 5-year relative survival rates for contemporary cancer patients. Specifically we were interested in the time interval between diagnosis and conditional relative survival becoming close to 100%, whether patients with specific cancers had a deteriorating long-term conditional survival, and in the long term influence of prognostic factors such as age and tumour stage. We considered conditional 5-year relative survival for every additional year survived up to 20 years, using population-based data from the long-standing Eindhoven Cancer Registry.

patients and methods The Eindhoven Cancer Registry has collected data on all patients with newly diagnosed cancer in the southern part of the Netherlands since 1955 (completeness >95% [10]). The registry now serves a population of 2.3 million inhabitants. The area offers good access to specialized medical care in 10 general hospitals and two large radiotherapy institutes.

original article

Annals of Oncology

Data on vital status, available up to 1 January 2006, were obtained from the hospital records and the death register of the Central Bureau for Genealogy, which registers all deceased in the Netherlands via the municipal civil registries. Patients who had moved and died outside the Netherlands may have been wrongly considered as being alive, but the estimated proportion of these patients was 95% from year Point estimate Upper end of 95% CIa

13 12 20 18 14 13 10 16 15 12 13 15 20 9 14 20 20 20 20 18 11 9 5 13 15

c c

12 6–10 3 10 7 9 11 c c

c

7 3 6 7 c c

15 c c c c c c c

5 7 6 5 2 5 6 6 8 10 5 0 5 3 17 15 9 15 17 11 9 c

3 0

95% CI = 95% confidence interval. a Upper end of 95% CI includes 95% from year. b Standard error of conditional survival proportion >5%. c Conditional 5-year relative survival proportion of >95% not reached within available follow-up period with reliable estimates for conditional survival.

Volume 18 | No. 8 | August 2007

doi:10.1093/annonc/mdm127 | 1409

original article

Annals of Oncology

Figure 1. Conditional 5-year relative survival rates for each additional year survived for cancer patients (M+ excluded), and 5-year relative survival at diagnosis. (a) Stomach, according to age; (b) colon, according to age; (c) colon, according to stage; (d) rectum, according to age; (e) NSCLC, according to age; (f) cutaneous melanoma males, according to age; (g) cutaneous melanoma females, according to age; (h) breast cancer, according to age; (j) breast cancer, according to stage; (k) prostate cancer age 50–74 years; (l) prostate cancer age 50–74 years, according to stage; (m) Hodgkin lymphoma age 25–49 years, according to gender.

1410 | Janssen-Heijnen et al.

Volume 18 | No. 8 | August 2007

Annals of Oncology

original article

Figure 1. Continued.

Volume 18 | No. 8 | August 2007

doi:10.1093/annonc/mdm127 | 1411

original article SAS computer package (SAS Institute Inc., Cary, NC, USA, 1999), using a publicly available macro for period analysis [14].

results Table 1 shows the numbers of patients available for conditional period survival analysis after 5, 10, 15 and 20 years for each tumour type, the number of years after diagnosis when there was hardly any excess mortality (conditional 5-year relative survival is consistently >95%) and the number of years after diagnosis when the upper end of the 95% confidence interval includes 95% conditional 5-year relative survival. Figure 1 shows the relative survival curve at diagnosis for each tumour type, as well as the conditional 5-year relative survival rate for each additional year survived. Five-year relative survival at diagnosis for patients with stomach cancer (M+ excluded) was rather poor (41% for those aged 25–49 years and 31% for those aged 50–74 years). However, the chance of surviving another 5 years increased steeply to 89% and 83%, respectively, for those still alive after 5 years (Figure 1a). For patients with colon or rectal cancer, conditional 5-year relative survival also increased with time since diagnosis, although much less steeply. For those aged 25–49 years, conditional 5-year relative survival was > 95% after 12 and 9 years for colon and rectal cancer, respectively (Figure 1b and d). The same pattern was seen for those aged 50–74 years, although conditional relative survival for the elderly with colon cancer slightly decreased again after 10 years. Although survival at diagnosis was clearly different for each stage group, this difference (almost) disappeared after survival for 6 years (Figure 1c). For patients with NSCLC, conditional 5-year relative survival never exceeded 91% for those aged 25–49 years and 77% for those aged 50–74 years (Figure 1e). Five-year relative survival at diagnosis was 70–80% for male patients with cutaneous melanoma in contrast to 89–93% for female patients. Conditional survival was very similar for the two age groups, and after 3–7 years, conditional 5-year relative survival was almost equal to survival in the general population with the same age structure (Figure 1f and g). For patients with breast cancer, 5-year relative survival at diagnosis was 84% for both age groups (Figure 1h). Conditional 5-year relative survival became 90–94% between 8 and 20 years after diagnosis. For patients with stage I breast cancer, 5-year conditional survival was >95% at 15 years after diagnosis (Figure 1j). For patients with stage II or III disease 5-year conditional survival did not exceed 88% (18–20 years after diagnosis). For patients with prostate cancer conditional 5-year relative survival decreased from 89% at diagnosis to 81% 6 years after diagnosis, whereafter it increased again to 91% after 11 years (Figure 1k). This pattern was found for tumours of both stage I/II and stage III (Figure 1l). Five-year relative survival with Hodgkin lymphoma aged 25–49 years was 84% for male patients and 94% for female patients at diagnosis. For male patients conditional survival improved only slightly with each additional year survived, to

1412 | Janssen-Heijnen et al.

Annals of Oncology

96% at 10 years after diagnosis, but thereafter decreased again slightly (Figure 1m). For female patients conditional survival remained more or less stable over time. There were not enough patients to reliably estimate conditional survival rates for the age group 50–74 years.

discussion Patients have hardly any excess mortality when conditional 5-year relative survival exceeds 95%. Survival for this group of patients is then similar to the general population with the same age structure. This was found for patients with colorectal cancer, cutaneous melanoma or stage I breast cancer. However, for other tumours (stomach cancer, NSCLC, stage II or III breast cancer, prostate cancer or Hodgkin lymphomas), conditional 5-year relative survival did not exceed 75–94%. This means that, even when alive at 10–20 years after diagnosis, patients with these tumours still have a poorer survival compared with the general population with the same age structure [15-17]. Potential explanations for the poorer survival compared with the general population include late recurrences, secondary tumours, or a higher death rate due to co-morbidity associated with the risk factors for cancer (e.g. chronic obstructive pulmonary disease and cardiovascular diseases, which are related to smoking in patients with lung cancer). For some tumour types (such as Hodgkin lymphomas or breast cancer), conditional 5-year relative survival decreased again slightly for patients who had already survived for some time. This could be explained partly by recurrence or progression even after long follow-up [1, 15]. Unfortunately we do not have these data available in our population-based cancer registry. Conditional survival can also decrease after some time because of late side-effects of treatment, such as pulmonary or cardiac conditions or second tumours (e.g. after Hodgkin lymphoma or breast cancer) [18-21]. For prostate cancer, conditional 5-year relative survival decreased up to 6 years after diagnosis, before increasing. The reason for this phenomenon is not clear. An increase in mortality after some time might be expected in the case of a ‘wait and see’ policy or hormonal treatment. In patients up to 70 years this was the case in 30–35% of those with a T1 tumour and in 20–25% of those with a T2 tumour. However, we also found a decrease in conditional survival in the first years after diagnosis for patients with a stage III tumour. For most tumour types 5-year relative survival at diagnosis was somewhat more favourable for younger compared with older patients [22]. Previous studies have shown that age still influenced conditional survival up to 5 years after diagnosis [1, 4, 5]. In our study we also found a prognostic influence of age on conditional survival in the first years after diagnosis, but this influence largely disappeared after survival for 5–10 years for most tumour types, except for NSCLC. Five-year relative survival at diagnosis was also clearly different for the stage groups. Previous studies among patients with colon cancer, breast cancer or lung cancer (SEER database) still alive up to 5 years after diagnosis (up to 2 years for lung cancer), have shown that the difference in survival at diagnosis between stage groups decreased with

Volume 18 | No. 8 | August 2007

Annals of Oncology

time since diagnosis. Stage, however, remained an important prognostic factor, even for conditional survival after 2–5 years [3, 5, 6]. In studies with a longer follow-up among patients with colon cancer (conditional survival rates up to 20 years after diagnosis), lung cancer (up to 5 years after diagnosis), cutaneous melanoma (up to 15 years after diagnosis) or breast cancer (up to 5 years after diagnosis), the difference between the stage groups decreased with time since diagnosis, or even disappeared [1, 8, 9, 15]. In our study we also found such a decrease in the differences between the stage groups.

conclusion Prognosis for patients with cancer generally improved with each year survived. The good news is that it improves for patients with most tumour types. Patients with colorectal cancer, cutaneous melanoma or stage I breast cancer exhibited hardly any excess mortality after 3–15 years, whereas for patients with other tumours survival remained poorer than for the general population. Insight into conditional survival is especially useful for (ex)patients, who may use this information to plan their remaining life.

acknowledgements This work was carried out with grants from the Dutch Cancer Society (IKZ 2000-2260) and the Muntendam Award of 2005 to JWWC.

references 1. Skuladottir H, Olsen JH. Conditional survival of patients with the four major histologic subgroups of lung cancer in Denmark. J Clin Oncol 2003; 21: 3035–3040. 2. Donovan RJ, Carter OB, Byrne MJ. People’s perceptions of cancer survivability: implications for oncologists. Lancet Oncol 2006; 7: 668–675. 3. Henson DE, Ries LA, Carriaga MT. Conditional survival of 56,268 patients with breast cancer. Cancer 1995; 76: 237–242. 4. Kato I, Severson RK, Schwartz AG. Conditional median survival of patients with advanced carcinoma: surveillance, epidemiology, and end results data. Cancer 2001; 92: 2211–2219.

Volume 18 | No. 8 | August 2007

original article 5. Merrill RM, Henson DE, Barnes M. Conditional survival among patients with carcinoma of the lung. Chest 1999; 116: 697–703. 6. Merrill RM, Henson DE, Ries LA. Conditional survival estimates in 34,963 patients with invasive carcinoma of the colon. Dis Colon Rectum 1998; 41: 1097–1106. 7. Punyko JA, Mertens AC, Baker KS et al. Long-term survival probabilities for childhood rhabdomyosarcoma. A population-based evaluation. Cancer 2005; 103: 1475–1483. 8. Hankey BF, Steinhorn SC. Long-term patient survival for some of the more frequently occurring cancers. Cancer 1982; 50: 1904–1912. 9. Gamel JW, George SL, Edwards MJ, Seigler HF. The long-term clinical course of patients with cutaneous melanoma. Cancer 2002; 95: 1286–1293. 10. Schouten LJ, Hoppener P, van den Brandt PA et al. Completeness of cancer registration in Limburg, The Netherlands. Int J Epidemiol 1993; 22: 369–376. 11. Houterman S, Janssen-Heijnen ML, van de Poll-Franse LV et al. Higher longterm cancer survival rates in southeastern Netherlands using up-to-date period analysis. Ann Oncol 2006; 17: 709–712. 12. Brenner H, Gefeller O. An alternative approach to monitoring cancer patient survival. Cancer 1996; 78: 2004–2010. 13. Brenner H, Hakulinen T. Up-to-date long-term survival curves of patients with cancer by period analysis. J Clin Oncol 2002; 20: 826–832. 14. Brenner H, Gefeller O, Hakulinen T. A computer program for period analysis of cancer patient survival. Eur J Cancer 2002; 38: 690–695. 15. Hatteville L, Mahe C, Hill C. Prediction of the long-term survival in breast cancer patients according to the present oncological status. Stat Med 2002; 21: 2345–2354. 16. Louwman WJ, Klokman WJ, Coebergh JW. Excess mortality from breast cancer 20 years after diagnosis when life expectancy is normal. Br J Cancer 2001; 84: 700–703. 17. Zahl PH, Tretli S. Long-term survival of breast cancer in Norway by age and clinical stage. Stat Med 1997; 16: 1435–1449. 18. van Spronsen DJ, Post PN, Crommelin MA et al. Modest decline in late mortality following Hodgkin’s disease in the southeastern Netherlands since 1972. Ann Hematol 1998; 76: 205–209. 19. Soerjomataram I, Louwman WJ, Lemmens VE et al. Risks of second primary breast and urogenital cancer following female breast cancer in the south of The Netherlands, 1972–2001. Eur J Cancer 2005; 41: 2331–2337. 20. Lorigan P, Radford J, Howell A, Thatcher N. Lung cancer after treatment for Hodgkin’s lymphoma: a systematic review. Lancet Oncol 2005; 6: 773–779. 21. Hill DA, Gilbert E, Dores GM et al. Breast cancer risk following radiotherapy for Hodgkin lymphoma: modification by other risk factors. Blood 2005; 106: 3358–3365. 22. Janssen-Heijnen ML, Houterman S, Lemmens VE et al. Prognostic impact of increasing age and co-morbidity in cancer patients: a population-based approach. Crit Rev Oncol Hematol 2005; 55: 231–240.

doi:10.1093/annonc/mdm127 | 1413