Long-term survival of cancer patients compared to heart failure and stroke: A systematic review

The facts that multiple diseases are present in many patients and that this trend is expected to increase in the future due to population ageing reveal the necessity for a better understanding of competing risks. Among the diseases with high mortality cancer, heart failure and stroke represent major global healthcare problems.

Cancer is the second most common cause of death after cardiovascular diseases. According to the World Health Organization (WHO), more than 10 million people are diagnosed with cancer yearly. The disease is responsible for 6 million deaths per year accounting for up to 12% of all cases. Fifty six percent of newly diagnosed cancer patients are >65 years, while about 70% of cancer deaths are in this age group. The median age of cancer patients at death for both sexes ranges from 71 to 77 years [1]. The four most common malignancies in developed countries are lung, breast, prostate, and colorectal cancer. These account for nearly half of all incident cases and cancer deaths of the total European cancer burden [2]. The most common entity overall and the leading cause of cancer related mortality is lung cancer. Worldwide about 1.35 million new cases and about 1.18 million deaths are estimated annually [3]. Among women the most common entity in developed countries is breast cancer. The disease is diagnosed in about 1.2 million patients and accounts for about 500,000 deaths yearly in the world [4]. Prostate cancer represents the most common cancer in men in developed countries with estimations for 2007 revealing about 782,600 new cases and 254,000 deaths [5, 6]. Finally colorectal cancer is the third most common malignancy worldwide with the WHO estimating about 945,000 new cases and 492,000 deaths annually [7].

Heart failure is a growing cause of morbidity and mortality among cardiovascular diseases, with increasing prevalence in recent years [8]. Despite significant improvements, the disease continues to represent an enormous clinical challenge. Currently, about 5 million patients are suffering in the USA from heart failure, while more than 550,000 are diagnosed yearly [9]. Incidence of the disease approaches 10 per 1000 population after age 65 and mortality was about 300,000 cases in 2006 in the USA [10]. About 80% of patients with new diagnosed heart failure are >65 years old and 50% are >75 years old. The disease is currently one of the most frequent causes of hospitalization, with the annual number in the USA estimated to be over 1 million [11].

The third leading cause of death in developed countries represents cerebrovascular disease. In 2002, stroke was the cause of 5.5 million deaths worldwide, accounting for about 10% of total deaths. The prevalence of stroke in the USA is over 700,000 cases per year [12]. About 75% of all first stroke events occur after the age of 65 years [13]. The disease is associated with high mortality in the acute phase [14]. The annual risk of death after stroke is about 10% [15, 16], while the annual recurrence risk is 5% per year, similar to that seen in patients with coronary events [17].

Aim of the present article is to provide an overview of long-term survival of cancer, heart failure and stroke patients, based on the results of large locoregional and international studies. The choice of those diseases in our analysis is based on the epidemiological data concerning incidence, prevalence and mortality that underline the high impact of cancer, heart failure and stroke in public health and show that the diseases tend to affect patients of similar age.

Among diseases responsible for the majority of deaths worldwide, cancer, heart failure and stroke possess leading positions. Aim of this work is to provide an overview of long term survival of those diseases as described in large population- or hospital based studies from different developed countries. We chose cancer, heart failure and stroke for our analysis mainly because of their epidemiological characteristics. Incidence and mortality of the diseases demonstrate that they represent major global healthcare problems tending to affect patients of similar age. Furthermore, our clinical empiricism shows that in some cases the prognostic impact of each disease is unclear and that cancer is often considered to be the disease with the worst prognosis.

Most identified studies investigated 5-year observed survival for the disease of interest. In one of the largest population based studies, observed 5-year survival after diagnosis was about 43% for all cancer diseases [21]. Observed 5-year survival for heart failure varied between 26% and 52% and observed survival after a stroke event varied between 40% and 68%. However, comparison of observed survival has major limitations. The observed mortality rate within the cohort of interest does not represent the mortality rate associated only with the disease of interest, but is equal to the all-cause mortality rate in the reference population plus the excess mortality rate associated with the disease. Comparisons based on observed survival rates between different studies are not reliable, since they are affected by demographic differences between the different populations. In order to eliminate such effects calculation of relative survival rates is more appropriate. Relative survival rate is defined as the ratio of the observed survival rate in a group of patients to the survival rate expected in a group of people in the general population, who are similar to the patients with respect to all of the possible factors affecting survival at the beginning of the period, except for the disease under study [42]. A relative survival rate of 1 indicates that the mortality of the study group does not differ from that expected in the general population and that the mortality attributable to the disease is zero. For the calculation of relative survival rates the expected survival is usually estimated from nationwide population life tables, stratified by age, sex and calendar time. The major advantage of relative survival is that the information on cause of death is not required [43]. The analysis of the cancer related studies identified in our investigation demonstrated 5-year relative survival rates of 50% to 57% for all cancer entities. Differencies in 5-year relative survival was demonstrated for the different cancer entities. In particular, breast and prostate cancer showed 5-year relative survival rates of 73% to 89% and 50% to 99%, respectively, while lung cancer showed considerably lower survival rates (12% to about 18%). This result highlights the need to educate the patients and the general population that "cancer" is not one disease, but an umbrella term for a number of malignancies characterized by tissue infiltration and metastatic dissemination but manifold symptomatology, varying response to treatment strategies, and different long-term prognosis.

Five-year relative survival rates for heart failure and stroke in the studies identified in our search was about 62% and 50% respectively. A comparison to the relative survival rates of the different cancer entities indicates that cancer might not necessarily have a worse prognosis. However, this conclusion is strongly limited by various factors. The most important limiting parameter is the lack of appropriate data in regard to relative survival ratios for heart failure and stroke. Our search identified only two studies for each disease, which is not enough for safe conclusions. Furthermore, the use of relative survival rates underlies the assumption that patients are subject to two independent forces of mortality, i.e. that attributable to the disease and that in the general population. However, the disease of interest is often included in the reference population, resulting in bias of the estimation. Cancer and stroke are more closely linked to a hospital admission, while the development of screening techniques, such as PSA-screening for prostate cancer and mammography for breast cancer, has led to an earlier diagnosis of the diseases and therefore a better identification of patients and discrimination from the reference population. In case of heart failure, however, many patients are first diagnosed when symptoms become severe enough to require hospitalization. The patients that are not recognized are possibly included in the reference population, resulting in bias of the estimated relative survival rates.

Past studies have directly compared heart failure vs. cancer-related survival within the same population, showing results that are similar to the trend identified in our comparison. A study performed by Stewart et al investigated all patients with a first admission to any Scottish hospital in 1991 for heart failure and the four most common types of cancer, revealing that with the notable exception of lung cancer, heart failure was associated with a worse survival rate compared to other common cancer entities [44].

Although comparison and analysis of long term survival of different diseases is limited, it still can be an important tool for a better understanding of competing risks. The competing risks concept describes the analysis of how mortality trends of one disease might influence the mortality trends of another disease. To analyse this different models are available. The model of Chiang eliminates a specific cause in order to estimate the effect on mortality from other causes [45], while the model of Rothenberg is based on the assumption that mortality from competing causes remains constant [46]. The influence of cerebrovascular and cardiovascular disease mortality trends on cancer mortality trends has been analysed in various studies. Llorca et al analysed the interrelation between cerebrovascular disease, ischemic heart disease and cancer mortalities in Spanish women in 1981 and 1994 using both models and showed that although cerebrovascular and ischemic heart disease mortality have decreased in all age groups during the investigation period, this had not a significant impact on cancer mortality [47]. However, investigation of competing risks usually assumes independence between the different causes of death. Different causes of death are considered to be independent when they do not share common risk factors. A prominent example of a common risk factor between cardiovascular, cerebrovascular disease and various cancer entities is tobacco smoking. The error that is produced by the assumption of independence between those diseases is therefore influenced on the smoking prevalence of the investigated population and on the level of mortality from tobacco-related cancers [47]. The results of our systematic review cannot provide safe conlusions on competing risks, since they compare long term survival of different populations and do not provide information on risk factors that are important for analysis. Still, the competing risks problem shows the necessity for studies investigating outcome of different diseases, especially considering the fact that the number of multimorbid patients is expected to increase in the future mainly due to population ageing.

In respect to survival trends, the studies analysed here reveal results that are moderately encouraging. Comparison of long-term survival shows that the prognosis of cancer and heart failure has modestly improved over the last decades. However, a better understanding of the underlying molecular mechanisms of the diseases and their pathophysiology, leading to the development of new diagnostic and screening methods and to innovative therapeutic strategies, is still needed.

The comparison made here has some limitations. One is the lack of appropriate data in regard to relative survival rates of heart failure and stroke. Furthermore, most population-based studies use administrative data from registries and therefore miss detailed clinical information and stage-specific analysis. Retrospective studies with observational design are limited by inaccuracies in the diagnosis and coding of the diseases. The results of the analysis may be influenced by incomplete information concerning prior diagnosis, stage, and severity of the disease and variations in diagnostic accuracy and therapeutic approaches. Competing health risks of patients in clinical studies are not always fully analyzed because the final cause of death is not precisely reported.

The analysis presented in this paper provides indications that cancer does not necessarily have a poorer prognosis compared to other common causes of death such as heart failure or stroke. Considering the fact that multiple diseases are present in many patients and that this trend is expected to increase in the future due to population ageing, there is an emerged need for a systematic investigation and statistical analysis of long term survival of different diseases within a population, coordinated by interdisciplinary teams. In this way, the role of competing risks will be better evaluated, which is important in order to better understand individual prognostic factors in multimorbid patients, prioritize the treatment of the prognostically leading disease and optimize therapeutic outcome.