With the definitions for critical illness insurance under review, Daniel Ryan looks at how changes to prostate cancer may impact the market
In May 2002 the Association of British Insurers (ABI) updated its best practice definitions for cancer claims under critical illness (CI) policies. The new definitions excluded: 'All tumours of the prostate unless histologically classified as having a Gleason score greater than six or have progressed to at least TNM classification T2N0M0.' As some policyholders have already been affected by these changes, there are two important questions that need to be addressed:
• Is it reasonable to expect policyholders to understand the implications of the above exclusion on the validity of future claims?
• Why was this exclusion introduced, and will others follow?
The answer to the first question will ultimately depend on the opinion of the Financial Ombudsman Service (FOS). However, to understand what the changes have meant, a brief background to Gleason scores is necessary.
Prostate cancers are graded by pathologists according to the degree to which the normal glandular structure is maintained, with grade one showing a normal prostate and grade five showing no evidence of gland units. The Gleason score represents the sum of the two most commonly seen grades in a biopsy. In general, higher Gleason scores are associated with worse prognosis. A Swedish study in 2002 that followed 305 prostate cancer cases concluded that 58% of those diagnosed with Gleason scores four to six, survived for more than five years, compared to 33% of those with Gleason scores of seven to ten.
The search for site-specific biomarkers that detect early cancers, as opposed to providing information on the effect of treatment or the recurrence of cancer, has lasted two decades with little success. The rapidly developing fields of proteomics and DNA micro-arrays may change this in the future by detecting differences in proteins and gene expression between normal and cancer cells.
Early successes
Prostate specific antigen (PSA) represents one of the early successes, although there are issues with the specificity and sensitivity of the PSA test, as PSA levels may rise because of infection or benign prostate hypertrophy, and some prostate cancers are associated with low levels of PSA.
Variants on the basic test have been developed to reduce the number of false positives, such as change in PSA levels over time (PSA velocity), comparisons of the level of PSA to the size of the prostate (PSA density) and the percentage of PSA bound to other proteins. In the US, a PCA3 test, or prostate cancer antigen 3, is available, and studies suggest a high level of specificity (90%) even in cases with low PSA levels. A needle biopsy is required to make a diagnosis of prostate cancer, but the introduction of widespread PSA testing has led to a classification of T1cN0M0 where the cancer would not otherwise have been detected by touch or imaging.
Earlier detection of prostate cancer because of PSA testing will lead to apparent improvements in prognosis even in the absence of development in treatment, as survival rates are measured from the date of diagnosis. The slow development of most prostate cancers would mean that these T1cN0M0 cancers would not represent a significant threat to the individual's life expectancy.
A study by Guileyardo in 1980 indicated that approximately 35% of those over age 50 showed evidence of prostate cancer at autopsy, but the percentage of deaths where prostate cancer is stated to be the principal cause is far less.
In the US, where PSA testing is more prevalent than in the UK, the five year relative survival probability for prostate cancer has increased from 45% for diagnoses in 1973 to 99% for diagnoses between 1995 and 2000. In the UK the improvement has been less dramatic from 33% for diagnosis in 1971/75 to 54% for 1990/94 as determined in the EUROCARE-3 study.
The expectation must be that wider PSA testing in the UK will lead to increased prostate cancer detection and further improvements in five year relative survival rates through lead-time bias. The intended effect of the ABI's exclusion was no doubt to limit claims on CI policies to those prostate cancers with poorer prognosis, although it could be argued that such amendments could have been included earlier given the body of knowledge in the medical profession that has existed since the early 1980s with regard to PSA and the prevalence of asymptomatic prostate cancer.
Research by Partin and Walsh in the 1990s at Johns Hopkins and Baylor College of Medicine, suggests that the prognostic value of the Gleason score is improved if levels of PSA and the estimated clinical stage are considered.
The Partin tables developed from this research suggest that for cancers with an estimated clinical stage of T1cN0M0, those with a Gleason score of five or six but high levels of PSA (>10ng/ml) may have a worse prognosis than those with a Gleason score of seven but low levels of PSA (<4ng/ml). It would be difficult in the light of this evidence to conclude that claims in respect of the former category of cancers were valid but the latter were invalid.
Ambitious goal
The ABI's suggested exclusion for prostate cancer is potentially unfair to those on the boundary, but consistent with the 'all-or-nothing' concept of CI payments. Indeed, the common practice of 'wait and see' treatment for some prostate cancers, because of uncertainty over the natural history, may mean that claims for those diagnosed with Gleason score six would be delayed rather than avoided. However, now that the concept of restricting claims based on the severity of a cancer has been broached, there may be further exclusions in the future, since, as the table opposite demonstrates, improvements in cancer survival have not been limited to prostate cancer.
These improvements are not surprising. There are currently many different research pathways focusing on reducing the burden of cancer. Indeed, the National Cancer Institute in the US has an ambitious but clearly stated goal "of eliminating suffering and death from cancer by 2015."
It may not be possible to determine the extent to which advances will relate to the prevention and early detection of new cancers as opposed to treatment of existing cancer and prevention of recurrence. Only those advances that prevent new cancers would reduce cancer-related CI claims. Improvements in early detection of those cancers that are usually only present in an advanced state, such as lung and stomach cancer, and where the individual may die of other causes prior to such presentation, would be expected to lead to increased claims experience for standalone policies.
Any new exclusion expressed in technical medical terms runs the risk of either the FOS deciding that such exclusions could not be reasonably understood by policyholders or of further research questioning the precise formulation of the exclusion. Examples in the case of prostate cancer would include:
• Research at Stanford University Medical Centre that has suggested that a modified Gleason score that considered the percentage of cancer classified as Gleason grades four and five proved a better predictor of prognosis than the unadjusted Gleason score.
• Joint research by the Institute of Cancer Research and the University of Liverpool that has identified the gene E2F3 as a marker for how aggressive a prostate tumour will be. The E2F3 gene had been previously linked to bladder cancer, and has a control function on cell division. An over-active E2F3 leads to uncontrolled cell proliferation. The study indicated that 67% of prostate cancer cells showed a degree of staining indicating the over-active gene, with higher levels of staining being associated with a worse prognosis.
An alternative to an increasingly complex structure of different exclusions by cancer site could be to exclude cancer claims based on a more objective measure of severity, such as predicted life expectancy after diagnosis based on prior experience for similar cases. This approach might be more easily understood by policyholders than a plethora of medical terminology, and would not require constant revision in response to changes in medical practice.
However, it is probably simplistic to assume that the perceived impact of a cancer diagnosis to the individual can be related to the effect on future life expectancy. The personal circumstances and history of the individual will play a huge part in the individual's reaction to such a diagnosis and may have been important factors in the decision to purchase a CI policy in the first place.
Daniel Ryan is senior health consultant at Watson Wyatt
COVER notes
• Prostate cancer is graded by pathologists according to the degree to which the normal glandular structure is maintained, with grade one showing a normal prostate and grade five showing no evidence of gland units.
• Any new exclusion expressed in technical medical terms runs the risks of either the FOS deciding that such exclusions could not be reasonably understood by policyholders or of further research questioning the precise formulation of the exclusion.