Conditional five-year relative survival improved successively with each additional year that patients lived following diagnosis for stomach, colorectal and lung cancer. This pattern was also similar to the regional or distant metastasis cases. Breast and prostate cancer showed different trends; conditional five-year survival was stable at a higher level. Liver cancer did not show any increase in conditional five-year survival.
Stomach and colorectal cancer
For stomach and colorectal cancer patients who survived more than five years after diagnosis, conditional five-year survival was close to 100%. This means that those who survived five years would have the same survival probability as the general population, i.e. they can be considered as ‘cured’ of cancer. Although it is difficult to define clinical ‘cure’ at the individual level, we can define the concept of ‘cure’ at population level [
21,
22]. Conditional five-year survival for localised cases was stable at more than 90%. Those for regional or distant metastasis increased according to the number of years since diagnosis. Even late stage patients who survive a few years have a chance of living another five years.
Lung cancer
Conditional survival in lung cancer patients increased according to the additional years after diagnosis. However, the figures were lower than those for stomach and colorectal cancer. This was because lung cancer patients have a higher risk of death due to complications related to cancer or cancer risk (smoking), such as ischemic heart disease.
Breast and prostate cancer
Conditional five-year survival rates for all cases of both breast and prostate cancer were around 80-90%, due to the higher proportion of localised patients in all cases. Conditional five-year survival for localised prostate cancer patients slightly decreased five years after diagnosis. This could be partly explained by the recurrence or progression of tumours during long-term follow-up. For these cancers, we need to follow-up patients for a longer period.
Liver cancer
Conditional survival for liver cancer was much lower than for other cancers at any stage or age after several years. Even in localised patients, conditional five-year survival was less than 40% after five years. This is probably because many liver cancer patients experienced a recurrence of cancer, or died from liver cirrhosis or liver failure related to the hepatitis B or C virus.
Effect of age and stage at diagnosis
Trends in conditional survival by age group were quite similar except for liver and prostate cancer. For most cancers, age did not significantly affect conditional survival. In the case of liver cancer, conditional survival for young patients (15–49 years old) was higher than for old patients (60–69 years old) after several years. This could be explained by the fact that the old patients had been exposed to hepatitis viruses for long time; as a result, they tended to develop liver cirrhosis and liver failure more than young patients. Conditional survival of young prostate cancer patients (50–59 years old) was lower than old patients (60–69 years old). This is probably because young patients are diagnosed at a more advanced stage than old patients (the proportion of distant metastasis was 35.2% in 50-59-year-old patients and 28.3% in 60-69-year-old patients).
The conditional survival curve was different by stage; stage at diagnosis was an important prognostic factor. Conditional survival for localised patients was stable at 85-95%, while for regional and distant metastasis patients it increased after several years of diagnosis.
Trends in conditional survival for breast and colorectal cancer patients in Osaka were similar to other countries (shown in Additional file
1: Figures S1-S4 from Australia [
2,
6], US [
8], Canada [
5,
11] and European countries [
3,
9]). Conditional survival for prostate cancer at all stages in Osaka was lower than other countries. This is due to the low proportion of localised patients in Japan compared to other countries. Conditional survival of stomach cancer for all stage and localised in Osaka was higher than in Australia [
2]. Stomach cancer patients in Osaka were diagnosed at an earlier stage than in Australia (e.g. 51% patients diagnosed at localised stage in Osaka, 28% in Australia). In addition, approximately half of the stomach cancer patients in Japan were diagnosed at T1 (UICC TNM classification) [
23]. Therefore we can estimate a higher proportion of T1 in localised patients in Osaka than in Australia. Higher conditional survival for localised patients can be partly explained by differences in tumour. This may be due to stomach cancer screening programmes [
24] and wide use of endoscopy in clinical settings in Japan. Conditional survival of localised lung cancer in Osaka was higher than in other countries. This could be explained by differences in tumour size and histology [
25,
26]. Conditional survival for liver cancer patients in Canada increased some years after diagnosis [
11], while in Osaka it was stable at low survival. This can be explained by the differences in etiological factor among these countries. In the US and Canada, prevalence of hepatitis B or C viruses in liver cancer cases was lower than in Japan [
27]. Liver cancer patients in Japan might have greater likelihood of liver failure or hepatitis-related cirrhosis than those in the US and Canada.
Conditional five-year survival for stomach and colorectal cancer patients who were alive five years after diagnosis was about 100%; this means those patients have similar survival probability to the general population. Therefore those patients can be considered as ‘cured’. For other sites of cancer, further long-term follow-up time may be needed to estimate ‘cured’ time.
Conditional survival is an important statistic for planning long-term life after diagnosis, not only for cancer patients and their families, but also patients with other diseases. However, a population-based disease registry system, such as the population-based cancer registry, is essential to estimate this type of statistic.