Article Text
PDF
Abstract
Background Neurofibromatosis 1 (NF1), neurofibromatosis 2 (NF2), familial adenomatous polyposis (FAP), von Hippel-Lindau syndrome (VHL), and Gorlin syndrome (GS) are single gene diseases that predispose to early onset tumours. Few studies have assessed the effect of these diseases on life expectancy. This study's aim was to assess this effect, and to test the hypothesis that genetic registers increase survival.
Method NF1, NF2, VHL, FAP, and GS patients were identified through the North West Regional Genetic Register Service and the North West Cancer Intelligence Service. Information on benign and malignant tumours, and deaths were obtained. Kaplan–Meier curves were used to show actuarial survival rates for each disease, compared to the local population, and in patients diagnosed pre/post the regional genetic register. Log rank (Mantel–Cox) tests were used to compare survival between groups.
Results Life expectancies were significantly reduced for all diseases investigated compared with the local population (80.0 years) (p=0.05). GS had the longest life expectancy at 73.4 years, followed by NF1 at 71.5 years, NF2 at 69.0 years, FAP at 63.6 years, and VHL at 52.5 years. Patients diagnosed after establishment of the genetic register had an increase in survival compared to those diagnosed pre-1990: NF2 (14.7 years), FAP (13.9 years), VHL (16.3 years), and GS (11.2 years).
Conclusion Life expectancy for all five diseases was less than normal, although in recent years this reached the level of the local population in GS. Although there have been improvements in all conditions which may in part be attributable to better targeted care through the genetic register service, more needs to be done to address the very poor life expectancy in VHL.
- Survival
- neurofibromatosis
- von-Hippel-Lindau syndrome
- familial adenomatous polyposis
- Gorlin syndrome
- hereditary disease
- genetics
- epidemiology
- genetic epidemiology
- clinical genetics
- genetic screening/counselling
- cancer: breast
- prevention
- cancer: CNS
- cancer: colon
- genetic epidemiology
- oncology
Statistics from Altmetric.com
- Survival
- neurofibromatosis
- von-Hippel-Lindau syndrome
- familial adenomatous polyposis
- Gorlin syndrome
- hereditary disease
- genetics
- epidemiology
- genetic epidemiology
- clinical genetics
- genetic screening/counselling
- cancer: breast
- prevention
- cancer: CNS
- cancer: colon
- genetic epidemiology
- oncology
Introduction
It is well recognised that certain single gene diseases predispose to early onset tumours resulting in a shortened life expectancy; these include neurofibromatosis type 1 (NF1), neurofibromatosis type 2 (NF2), familial adenomatous polyposis (FAP), von Hippel-Lindau syndrome (VHL), and Gorlin syndrome (GS). In 1989, the Regional Genetic Service in Manchester established a genetic register to facilitate the long term management of a number of genetic conditions including those listed above.
NF1 is an autosomal dominant disease with an estimated birth incidence of 1 in 2699.1 Features include café-au-lait patches, and peripheral and plexiform neurofibromas.1 The risk of malignant tumour development in NF1 is much greater than the general population2–6; malignant peripheral nerve sheath tumours (MPNST), central nervous system tumours, and vasculopathy are the most frequent causes of premature death.7 The decrease in NF1 life expectancy was previously estimated as 15 years.8 9 However, a recent study by Evans et al10 found the reduction in life expectancy to be lower at only 8 years. FAP now has a similar life expectancy to NF1 (estimated 70.4 years)11 and is one of a group of inherited polyposis syndromes which predispose to colorectal cancer. If unscreened, >90% of FAP patients will develop colorectal cancer, which is therefore the main recorded cause of death.12 Extra-colonic features include malignancies of the upper intestinal tract, brain, thyroid, and hepatobiliary systems, and desmoid disease.
Previous studies have demonstrated that patients with NF2 have an even greater reduction in life expectancy, often caused by the pressure effects of benign tumours (such as peripheral and spinal root nerve schwannomas) associated with the disease.13–15 A study by Evans et al16 reported an average (mean) age at death of only 36 years. However, life expectancy is likely to be increasing due to better surgical management.12 Similarly, the average survival for patients with VHL is reportedly <50 years,17 18 with median actuarial survival reported at 49 years by Maher et al.19 VHL is characterised by the development of tumours in highly vascular areas; retinal angiomas and central nervous system haemangioblastomas are most common.19 20 While these tumours are benign, some patients develop such a significant burden of CNS haemangioblastomas that they lead to early death. There is also an increased risk of phaeochromocytomas and renal cell carcinoma (RCC) in patients with VHL, with RCC resulting in reduced life expectancy.
With an incidence of 1 in 18 000, GS is one of the more common cancer predisposing diseases and is characterised by the presence of multiple basal cell carcinomas (BCCs) and developmental anomalies such as a bifid ribs, macrocephaly with frontal bossing, and eye abnormalities.21 The recurrent BCCs develop from late adolescence, although metastases are rare.21 22 Approximately 5% of affected people develop medulloblastoma during early childhood.22 Despite the increase in risk of malignancy, life expectancy is not thought to be greatly reduced in this disease.21 23
Few studies have investigated life expectancies in these hereditary cancer predisposing diseases, particularly in relation to ‘all cause’ mortality. The majority of existing research has focused on the more common conditions of NF1 and FAP, and on highly specific situations such as survival post-surgery24 or survival from tumour development.25–27 Many reports consider 1 year or 5 year survival but do not cover extended periods,28 29 nor do they use large patient cohorts. This study therefore sought to calculate life expectancy of NF1, NF2, FAP, VHL, and GS patients who were highly ascertained in the Manchester region (as indicated by our recent assessments of disease prevalence that are among the highest published1), and compare these to the life expectancy of the general population.
We have previously published on life expectancy for NF1 from the regional genetic register1 10 and on the effects of screening and genotype on life expectancy in FAP.11 The register has allowed an estimate of birth incidence, disease prevalence, and de novo mutation rates. It is believed that genetic registers facilitate more rigorous screening, earlier diagnosis, and better follow-up and treatment which may then lead to an increase in survival.11 In order to test this hypothesis, we analysed life expectancy in patients diagnosed both before and after the establishment of the regional genetic register.
Patients and methods
The North West Regional Genetic Register Service (GR) was established in 1989 and includes information (demographic, pedigree, genetic, screening, and disease symptomatology) on individuals and families with genetic diseases such as NF1, NF2, FAP, VHL, and GS. The GR covers Greater Manchester and Lancashire (GML) but not Merseyside, a population of 4.1 million. The North West Cancer Intelligence Service (NWCIS) covers this same population and collects information on all patients resident in GML who develop cancer. A number of disease specific databases for register conditions have also been developed. This was a part prospective, part retrospective cohort study; only patients resident in GML were included in this study. Cases (proven or obligate gene carriers) were identified from the GR using existing diagnostic criteria for NF1,30 31 NF2,14 32 FAP,33 VHL,34 and GS.21 Their data were verified against medical records, NHS Summary Care Records, and the NWCIS. Disease specific databases were interrogated on 10 June 2010 and information was gathered on the status of the patient and date of birth. Data were censored at date of most recent follow-up (date the patient was last in contact with the department or other NHS service) or date of death (obtained from death certification either directly or via the NWCIS). Patients were excluded if there was no known date of death or date of last follow-up (seven FAP and two GS patients). Life expectancy data for the general North-West population was calculated using rates of death that had been determined by the North West Cancer Intelligence Service. Cause of death was determined where possible from medical records, NWCIS data or death certification; specific note was made when one of the cancer predisposing genetic diseases was recorded on the death certificate. Deaths were attributed to the disease if due to a clearly documented disease association. Actuarial survival rates for each disease group were calculated using Kaplan–Meier curves and compared to rates for the GML population. Log rank (Mantel–Cox) tests were used to compare survival between groups. Median life expectancy was reported for all groups.
A further actuarial survival was calculated with patients grouped according to their date of ascertainment/diagnosis either before establishment of the genetic register (in 1989/1990), or after. Patients with unknown date of ascertainment or incomplete data were excluded (11 NF2, 46 FAP, 12 VHL, 27 GS patients). As the date of diagnosis was not reliable for many NF1 patients (due often to uncertainty from patient notes, especially about diagnosis in childhood), only NF2, FAP, VHL, and GS patients were included in the second analysis.
A final analysis of expected versus observed deaths was carried out for the whole cohort for four decades commencing in 1970 using death rates for each age group from the Office of National Statistics; however, follow-up was commenced at 1 year of age to avoid the perinatal and neonatal deaths that would not be relevant to the tumour prone disorders as the conditions would not have been diagnosed at that stage.
Results
Within GML, 1085 cases of NF1, 113 cases of NF2, 379 cases of FAP, 68 of VHL, and 170 cases of GS were identified (table 1).
Cohort breakdown by gender of the five genetically inherited cancer predisposing diseases included in survival analyses
The cohort breakdown for the hereditary cancer predisposing diseases is presented in tables 1 and 2. Potential confounders include incomplete ascertainment; however, this is unlikely to be biased between diseases.
Birth cohort distribution and deaths for each cohort based at end of 2010
Survival
Figure 1 shows the survival from birth of each cancer predisposing disease compared with the life expectancy in GML. Survival was significantly reduced for all cancer predisposing diseases investigated in comparison to the general population (p=0.05) (figure 1).
Kaplan–Meier survival curve comparing life expectancy in people with neurofibromatosis type 1 (NF1), neurofibromatosis type (NF2), familial adenomatous polyposis (FAP), von Hippel-Lindau syndrome (VHL), Gorlin syndrome and the Greater Manchester and Lancashire (GML) population. This figure is produced in colour in the online journal; please visit the website to view the colour figure.
Median life expectancy in GML was 80 years for 2007 (95% CI 79.8 to 80.2). Of the cancer predisposing diseases, GS was found to have the longest life expectancy at 73.4 years (95% CI 64.0 to 82.8 years), followed by NF1 at 71.5 years (95% CI 68.5 to 74.6 years), NF2 at 69.0 years (95% CI 58.9 to 79.0 years), FAP at 63.6 years (95% CI 58.6 to 68.7 years), and VHL at 52.5 years (95% CI 41.7 to 63.3 years). The breakdown of deaths by birth cohort can be seen in table 2. This shows a greater proportion of deaths in more recent birth cohorts in VHL and NF2 compared to the other conditions. Mean age at death was 41.3 years (95% CI 37.3 to 45.3 years) for 51 combined (all diseases) patients dying before 1970; 42.4 years (95% CI 38.5 to 46.3 years) for 100 patients dying between 1970 and 1989, and 52.1 years (95% CI 49.3 to 54.9 years) for 177 patients dying between 1990 and 2011. The observed difference between age at death pre- and post-genetic register (1990) was statistically significant (p<0.0001). The major change in mean age was between the 1980s (39.8 years; 95% CI 35.2 to 44.4 years) and the 1990s (52.5 years; 95% CI 48.5 to 56.4 years) (p<0.0001).
The median life expectancy in GML was 78.0 years for males and 82.0 years for females (p<0.0001). When separated by gender, the life expectancy in the diseased populations was found to be significantly reduced in males compared with the GML general population in all five diseases (NF1 p<0.0001, NF2 p=0.0004, FAP p<0.0001, VHL p<0.0001, GS p=0.0123), and in females with NF1, NF2, FAP, and VHL (p<0.0001), but not in females with GS (p=0.0618) (figure 2). Differences in life expectancy between males and females within the five diseases was not significant (NF1 p=0.065, NF2 p=0.520, FAP p=0.402, VHL p=0.953, GS p=0.238).
Median life expectancy for males and females. This figure is produced in colour in the online journal; please visit the website to view the colour figure. FAP, familial adenomatous polyposis; NF1, neurofibromatosis type 1; NF2, neurofibromatosis type 2; VHL, von Hippel-Lindau syndrome.
Cause of death
There were 297 deaths in the cohort of 1815 cases. Cause of death from death certification was determined in 255 (86%) cases. Of those with NF1, only 63/113 (55%) cases with a documented cause of death stated NF1 or a well recognised complication of NF1 as the underlying cause (table 3). Of those with NF2, 66% (23/35) of deaths were determined to have a disease related cause. Seventy per cent (42/60) of FAP deaths were attributed to the disease, with 73% of VHL deaths recorded as having a disease related cause. Only 19% (6/32) of the deceased GS patients with a documented cause were attributed to GS (table 3). Suicides could not be definitively attributed to the disease. There was one suicide in patients with NF1 (aged 29 years) and NF2 (aged 45 years), although the NF2 case was severely affected with poor quality of life. There were also two suicides among the 60 known deaths in FAP. In total there were four suicides out of 297 deaths among this cohort (1.35%). This is not significantly higher than the expected suicide rate among the general population (908 per 100 000 deaths, 0.91%) (p=0.43). When broken down by sex, the proportion of male suicides (three) and female suicides (one) of the 297 deaths reported here did not significantly differ from the suicide rates of males (1400 per 100 000 deaths) and females (416 per 100 000 deaths) in the general population (p=0.568 and p=0.832 respectively). Only a small percentage (23.5%) of death certificates recorded hereditary disease information, with 33% of NF1 patients having a clear indicator of disease related deaths. Only 10% of GS patient deaths had GS registered on the death certificate. These patients may not have died as a direct result of the disease, but the disease should be recorded on the death certificate (table 3).
Cases with cause of death established from death certification
Survival pre- and post-genetic register
An improvement in life expectancy was seen in patients diagnosed after the establishment of the genetic register compared to those diagnosed previously (table 4).
Comparison between median life expectancies by diagnosis pre-/post-North West Genetic Register
Patients with NF2 lived for approximately 14.7 years longer when diagnosed after 1990 compared to those diagnosed pre-1985 (p=0.02). This was mirrored in the other hereditary diseases, with an increase in survival of approximately 13.9 years for FAP patients (p<0.0001) and 16.3 years for those with VHL (p=0.02) (table 4). Life expectancy for GS patients also significantly improved from 70.1 years (pre-1985) to 81.3 years (post-1990; p=0.04) (table 4). Comparing the lowest quintiles across the registry creation time points (pre-1970, 1970–1989, 1990+), mean age at death was significantly improved in this youngest age group from 18.3 years (1970–1989) to 24.8 years (1990+) (p=0.013). Unexpectedly, age at death pre-1970 was higher (27.9 years) than post-register 1990 (24.8 years) (p=0.04). This may be due to lack of ascertainment of non-familial cases that are generally more severe.
When compared with the median life expectancy of the North West general population in 2001 (80 years), the life expectancy of patients diagnosed after establishment of the genetic register remains significantly lower for NF2, FAP, and VHL patients (p<0.0001 for all). Conversely, life expectancy of GS patients diagnosed after 1990 was 81.3 years compared to 80.0 years in the North West general population, although this was not significant (p=0.80).
A general downward trend was found between observed and expected death rates from 1980 to 2010 (using the ONS crude death rates for 1978 (for the 1970s), 1988, 1998, and 2000). During the 1970s, those with a cancer predisposing disease had a threefold increased risk of death (RR 3.16) compared to the general population. This increased to an RR of 3.27 (54 observed, 16.5 expected deaths) in the 1980s, though a downward trend appeared after the introduction of the register in 1989, with the RR falling to 3.14 (84 observed, 26.7 expected deaths) in the 1990s, and 2.70 (110 observed and 40.7 expected deaths) in the 2000s. There were only 36 recorded deaths in the 1970s which is likely to mean that these were very substantially under ascertained.
Discussion
This study documents the life expectancy in five hereditary cancer predisposing diseases—NF1, NF2, FAP, VHL, and GS—in a large population and tests the hypothesis that the establishment of the genetic register improves survival.
GS had the longest life expectancy among the genetic disease patients at 73.4 years. However, survival was significantly reduced for all conditions investigated in comparison with the general population (80 years; p=0.05). The disease with the poorest life expectancy, VHL, should be a target for improved patient care. This poor prognosis may be due to the fact that despite early detection of disease associated tumours, some patients have such significant disease burden that either the sheer number and site of tumours in the case of CNS haemangioblastomas or complications of therapy (eg, renal dialysis/transplantation in those with multiple RCCs) still result in early death. New modalities such as radiofrequency ablation for the treatment of RCC may improve survival in this group over the next few years.35
Due to the systemic nature of tumours and their multiplicity in these syndromes, drug treatments such as those developed in NF236 37 are likely to be the most important new area for research. Although life expectancy in NF2 appears to be relatively good, this disguises an early peak of deaths in the 20s and early 30s. This is because <50% died young, but deaths until 40 years of age were equivalent to vHL. There appears to be a double curve with the mean life expectancy improved by the dominant second curve. The high level of ascertainment of mild mosaic patients in our region and the presence of a single, very large, mildly affected family38 may have biased the results of this study towards longer survival. Exclusion of mild mosaic cases and inclusion of only patients with typical NF2 mutations (present in all cells) would produce a lower overall life expectancy. The regional patients are also managed through a single multidisciplinary clinic which has previously been shown to be associated with better survival.15 39
Survival was found to be significantly reduced compared to the general population in males and females with NF1, NF2, FAP, and VHL (p<0.0001). Life expectancy in GS women was statistically not reduced (p=0.0618). The difference in survival between males with GS (70.3 years) and males in the general population (78 years) was significant (p=0.0123); however, this difference was the least significant of all five diseases. The data in the current study correspond with previous findings that patients with GS generally have a good prognostic outcome with near normal life expectancy23; this, however, is the first time that life expectancy in GS has been formally assessed. No survival difference was reported between genders by disease (p=0.238 to 0.953). On average, females had a slightly greater life expectancy for NF1, FAP, and GS compared to males with the disease, reflecting the trend of the general population. Conversely, women had a lower life expectancy compared to men when diagnosed with NF2 or VHL, though this difference was again non-significant.
Although death certificates are sometimes used as a way of overcoming ascertainment bias in cohort studies, they are completely dependent on the accurate recording of death and attributable cause of death certification.9 10 In this study only 63/113 (55%) of NF1 patients who had a recorded cause of death had NF1 as an obvious attributable cause (including secondary complications such as glioma and MPNST). Of these individuals, only 31/63 (49%) had NF1 specifically mentioned on the death certificate. We were able to obtain details of death certification for 15 VHL patients with 11 (73%) having died due to VHL complications. Once again a low number of patients (3/11, 27%) had their genetic condition recorded on the death certificate. This notable absence of hereditary cancer predisposing diseases on death certificates should be addressed. Under-representation of genetic disease attributable deaths could alter the calculated life expectancy in these patients.9 10 It also means that studies of cancer syndromes in the UK using death certification are likely to be highly inaccurate due to poor recording of any underlying genetic syndrome.9
There has been improved survival (11.2–16.3 years) among patients diagnosed with NF2, FAP, VHL, and GS since the establishment of the genetic register in 1989. This group has already reported the improvement in life expectancy associated with the genetic register in FAP,11 where organisation of screening and prevention of colorectal cancer is paramount. While there was only relatively weak support for this trend from an analysis of decade adjusted expected deaths among the register conditions, this is likely due to the fact that we were aware of nearly all deaths from 1990, but not of the deaths in particular that occurred in non-familial cases pre-1990. As such, the general downward trend in observed versus expected deaths adds support to the overall trend in life expectancy. It is reasonable to consider that this trend of increased survival in the other conditions may also be due to in part to improvements in patient awareness, as well as disease management attributable to the establishment of the North West genetic register, while acknowledging that some of this trend may be due to generalised improvement in life expectancy and to advances in screening and treatment over the period. Further improvements in these areas, combined with recent developments such as microsurgical removal, stereotactic radiosurgery, and even drug treatment,36 37 should ensure continued increases in survival. Although an obvious improvement in life expectancy can already be seen, the full benefits of the genetic register may take several generations to reach their full effect. It is likely that insufficient time has elapsed for optimum mortality reduction.
To our knowledge, the current study is the first to compare life expectancy in several tumour prone inherited conditions. The study was population based with high levels of patient ascertainment, particularly for affected individuals alive after 1990. The results of this study are generalisable, though different countries and areas within the UK will likely have less of a register based approach. The study may not have captured severely affected de novo cases without affected offspring who died before 1990; however, this would, if anything, increase the potential benefit of being diagnosed after the inception of the genetic register as these missed cases would have most likely died at younger ages. Support for the specific effect of the register is provided by the only significant rise in mean age at death from the five conditions between the 1980s and 1990s. We were not able to obtain death certification on all deaths and data were not complete on all patients, which may have caused minor changes in the results presented.
In conclusion, the current study has shown a reduction in life expectancy for all five cancer predisposing diseases compared to the general population, although in recent years this normalised in GS. Although there have been improvements in all conditions, which may in part be attributable to better targeted care through the genetic register service, more needs to be done to address the very poor life expectancy in VHL.
Acknowledgments
We acknowledge the National Specialist Commissioning team for their support of the National Complex NF1 service and the NF2 service. The department of Genetic Medicine in Manchester is supported by the NIHR Manchester Biomedical Research Centre. We acknowledge the statistical support of Dr Catherine O'Hara, North West Cancer Intelligence Service.
References
Footnotes
Funding This study was unfunded research. The sponsor of this study is Central Manchester Foundation Trust. This sponsor had no role in the study design, in the collection, analysis, and interpretation of the data, in the writing of this report, or in the decision to submit this report for publication. The guarantor for this submitted work is Professor Gareth Evans.
Competing interests None.
Provenance and peer review Not commissioned; externally peer reviewed.
Data Sharing Statement The authors agree to share these data.