Article Text

Download PDFPDF

Original Article
Risk factors for survival in patients with von Hippel-Lindau disease
  1. Jiang-Yi Wang1,2,3,
  2. Shuang-He Peng1,2,3,
  3. Teng Li1,2,3,
  4. Xiang-Hui Ning1,2,3,
  5. Sheng-Jie Liu1,2,3,
  6. Bao-An Hong1,2,3,
  7. Jia-Yuan Liu1,2,3,
  8. Peng-Jie Wu4,
  9. Bo-Wen Zhou1,2,3,
  10. Jing-Cheng Zhou1,2,3,
  11. Nie-Nie Qi1,2,3,
  12. Xiang Peng1,2,3,
  13. Jiu-Feng Zhang1,2,3,
  14. Kai-Fang Ma1,2,3,
  15. Lin Cai1,2,3,
  16. Kan Gong1,2,3
  1. 1 Department of Urology, Peking University First Hospital, Beijing, People’s Republic of China
  2. 2 Institute of Urology, Peking University, Beijing, People’s Republic of China
  3. 3 National Urological Cancer Center, Beijing, People’s Republic of China
  4. 4 Department of Urology, Beijing Hospital, Beijing, People’s Republic of China
  1. Correspondence to Professor Kan Gong, Department of Urology, Peking University First Hospital, Beijing 100034, People’s Republic of China; gongkan_pku{at}


Background Historically, von Hippel-Lindau (VHL) disease is characterised by a poor survival. Although genotype–phenotype correlation has been described in many studies, the risk factors for VHL survival remain unclear. This study aims to evaluate the median survival of Chinese patients with VHL disease and explore whether VHL survival is influenced by genetic and clinical factors.

Methods In this retrospective study, we recruited 340 patients from 127 VHL families. Kaplan-Meier plot and Cox regression model were used to evaluate the median survival and assess how survival was influenced by birth year, birth order, sex, family history, mutation type, onset age and first presenting symptom.

Results The estimated median life expectancy for Chinese patients with VHL disease was 62 years. Patients with early-onset age, positive family history and truncating mutation types had poorer overall and VHL-related survival. Patients with haemangioblastoma as their first presenting symptom were related to a higher risk of death from central nervous system haemangioblastoma than those with abdominal lesions (HR 8.84, 95% CI 2.04 to 38.37, P=0.004).

Conclusions This largest VHL survival analysis indicates that onset age, family history, mutation type and first presenting symptom have an effect on the survival of patients with VHL disease, which is helpful to genetic counselling and clinical decision-making.

  • von Hippel-Lindau disease
  • survival
  • risk factor
  • genetic cancer syndrome

Statistics from


von Hippel-Lindau (VHL) disease (OMIM 193300) is an autosomal-dominant hereditary cancer syndrome caused by germline mutation of VHL tumour suppressor gene.1 2 It is a rare monogenic disease with birth incidence of about 1 in 36 000–53 000 all over the world and has a high penetrance of >90% by 70 years.3–7Patients are predisposed to various types of tumours in their early ages, including central nervous system haemangioblastoma (CHB), clear cell renal cell carcinoma (RCC), retinal haemangioblastoma (RA), pheochromocytoma (PHEO), pancreatic cystic lesion (PCL), pancreatic neuroendocrine tumour (PNET), endolymphatic sac tumour, epididymal and broad ligament cystadenoma.1 8–10 Genotype–phenotype correlations have been described in many studies. VHL disease has been classified into two clinical subtypes: type 1 patients usually confer a lower risk for pheochromocytomas and harbour truncating mutations, while type 2 patients are characterised by missense mutations and an increased risk for pheochromocytomas.11

Historically, life expectancy of patients with VHL disease is significantly reduced compared with the general population. In the largest study of VHL survival to date, the median survival of patients with VHL disease was reported to be 49 years, but the conclusion was drawn in 1990.12 Recently, some other studies in selected VHL cohort (VHL-related PHEO or specific mutation types) reported the improved life expectancy of up to 64 years.13 14 Moreover, a retrospective VHL study including all known Danish VHL families confirms that the median survivals for male and female patients with VHL disease have reached 67 and 60 years, respectively. However, the survival of VHL in other countries across the world remains unknown.

Patients with VHL disease will develop tumours in multiple organs, and the most common causes of death are CHB (25%–51%) and RCC (13%–47%).15–20 Genotype has been confirmed to be related to the risk of tumours in specific sites, but its effect on patients’ survival is yet unclear. Previous studies reported that VHL survival had improved in patients born in later years, and surveillance attendance reduced death rates in truncating mutation carriers.21 However, the influences of other factors such as sex, onset age, family history, birth order and first presenting symptom on overall and VHL-related survival have not been evaluated or verified until now.

In this study, we aim to assess the overall and VHL-related survival of Chinese patients with VHL disease based on a large VHL cohort and determine how survival is affected by sex, family history, genotype, birth year, birth order, onset age and first presenting symptom. This is the largest study of VHL survival across the world and will provide important information to genetic counselling and clinical decision-making.

Patients and methods

Patient ascertainment and assessment

In this retrospective cohort study, we recruited all the patients with VHL disease diagnosed at Peking University First Hospital (the only International VHLA Clinical Care Center in China). Individuals were identified as patients with VHL disease if they carried a germline VHL gene mutation (n=244) or fulfilled the clinical criteria (n=131).10 All the 131 patients diagnosed with clinical symptoms had at least one affected relative identified by VHL mutation test, so that their genotype could be predicted. A total of 375 patients from 134 VHL families were diagnosed. Clinical characteristics including birth year, birth order, family history, mutation type, onset age and first symptom were obtained through interviews with living family members and review of medical records. Positive family history was identified when a parent was diagnosed as VHL disease (including mosaicism carriers) or as an obligate mutation carrier (without VHL symptoms but having more than one affected offspring). Onset age was defined as the age when the first symptom or sign occurred. Thirty-five patients were excluded because of obscure clinical information. Therefore, 340 patients with VHL disease from 127 families were enrolled for survival analysis.

Of the 340 patients, 88.8% (302/340) were affected by at least one VHL-related lesion and the rest were asymptomatic mutation carriers. Patients’ entire lifetimes from birth until death or the end of follow-up in June 2017 were included for analysis, and the median follow-up time was 37.5 years/person (range 1–75 years) with a total 13 140 person-years. Any patient who was affected by PHEO before death or the end of follow-up was grouped into type 2 VHL, while the others were defined as type 1 VHL.

Genetic analysis

At least one patient from every family received VHL mutation test. Genomic DNA was extracted from peripheral blood of suspected individuals using QIAamp DNA Blood Mini Kit (QIAGEN, Germany) according to instructions. Three coding exons and flanking intronic regions were amplified by PCR using primers as described before.22 Direct sequencing was performed to detect missense mutations, splicing mutations and small indels. For large deletions and duplications, multiplex ligation-dependent probe amplification P016-C2 kit (MRC-Holland, Amsterdam, the Netherlands) was used. All large exon deletions in this study were verified by real-time PCR with primers described by Ebenazer et al.23 The 340 patients with VHL disease were divided into missense group (missense mutations, n=165) and truncating group (nonsense, small indels, large deletions, splicing mutations, n=175), based on the integrity of mutational protein (online supplementary table 1).

Supplementary Material

Supplementary Table 1

Statistical analysis

The 340 patients with VHL disease were included into overall survival, VHL-related survival and CHB-specific survival analyses using Kaplan-Meier curve in different groups of sex, birth order, family history and mutation type. After excluding 38 asymptomatic mutation carriers and 4 patients without clear onset age, we evaluated the effect of onset age on overall survival and VHL-specific survival among 298 patients. Patients were divided into earlier-onset group and later-onset group based on the mean onset age (about 30 years). The first presenting symptom could not be confirmed in 9 of the 298 patients, and the rest of the 289 patients were then enrolled to analyse the influence of first symptom on survival.

Cox regression model was used to assess the effect of sex, birth year, birth order, family history, mutation type, onset age and first symptom on hazard of death (from any cause, VHL-related cause or CHB only). Onset age was treated as a continuous variable and a categorical variable (30 years as the cut-off value). In the first symptom analysis, CHB and RA were combined as haemangioblastoma (HB) group, for the same pathological type and for the reason that retina is a part of the central nervous system. RCC, PHEO, PCL and PNET were combined as abdominal group because they were characterised by other pathological types and sometimes two or three of them were found at the same time through ultrasound or CT. Then, all the seven factors were put into a multivariate Cox regression model to evaluate whether they are independent risk factors for overall survival, VHL-related survival and CHB-specific survival.

P value <0.05 is regarded as statistically significant in this study.


Clinical characteristics of Chinese patients with VHL disease

In this cohort, 13.5% (46 of 340) patients with VHL disease were identified as type 2 disease with the existence of PHEO, and the others were type 1. Among the symptomatic individuals included for analysis, the mean onset age was 31.2±12.8 years. In all the patients, 270 individuals had a positive family history, and 70 patients carried probable de novo mutation. 48.5% (165 of 340) of the patients were missense mutation, while the rest were truncating mutation. CHB was the first presenting VHL-related lesion in their lifetime for more than half of patients, followed by RCC, RA, PHEO and PCL. VHL disease displayed high age-related penetrance in China (figure 1A). The cumulative penetrance by 60 years was about 97%.

Figure 1

Age-related penetrance and survival of Chinese patients with von Hippel-Lindau disease. Kaplan-Meier curve was used to describe the penetrance (A) and overall survival (B). Log-rank test was performed to compare the difference between male and female patients (C) and different clinical types (D).

In this cohort, 72 out of 340 patients died. The mean age at death was 43.7±14.4 years, and 97.2% (70 of 72) of the causes leading to death were VHL-related, in which CHB (67.7%, 48 of 72) and RCC (27.8%, 20 of 72) were the most common causes (table 1). One of the two non-VHL-related deaths was suicide because of emotional problems, and another one was traffic accident.

Table 1

Characteristics of the patients included in the survival analysis

Survival of patients with VHL disease

The median survival for Chinese patients with VHL disease was 62 years (figure 1B). Women seemed to have a longer survival than men (69 vs 62 years), but the difference was not significant (P=0.294) (figure 1C). The patients with type 1 VHL disease showed a significantly poorer survival than patients with type 2 VHL disease (median survival 62 vs 72 years, P=0.016) (figure 1D).

Risk factors for overall survival and VHL-related survival

In this study, the median survival for patients with onset age before 30 years was 60 years, which was 6 years shorter than that for patients with onset age after 30 years (P<0.001) (figure 2A). Accordingly, the Cox regression analysis demonstrated that onset age had a highly significant effect on survival; the later the onset age was, the lower was the hazard of death (HR 0.96, 95% CI 0.94 to 0.98, P<0.001). In a multivariate Cox regression model, the risk of death in earlier-onset group (onset age ≤30 years) was almost three times than the later-onset group (HR 2.70, 95% CI 1.46 to 4.99, P=0.002) (table 2). The effect of onset age was similar when we considered risk of death only from VHL-related reason (HR 2.41, 95% CI 1.29 to 4.50, P=0.006) (online supplementary table 2).

Supplementary Material

Supplementary Table 2
Figure 2

The effects of different factors on overall survival of patients with von Hippel-Lindau disease. Kaplan-Meier curve and log-rank test were used to compare the differences between patient groups by onset age (A), family history (B), mutation type (C) and first presenting symptom (D).

Table 2

Univariate and multivariate analyses for risk factors of overall survival

The overall survival of patients with VHL disease depended on family history and genotype as well. Patients with positive family history had a median survival of 61 years, while the median survival was 69 years for those without family history (P=0.005) (figure 2B). As to genotype, patients with missense mutations had a more favourable survival than truncating mutation carriers (median survival: 66 vs 60 years, P=0.045) (figure 2C). Additionally, multivariate Cox regression model showed that positive family history and truncating mutation type were independent risk factors for poor survival (table 2). The effect of family history was similar if only the risk of VHL-related death was considered (HR 2.08, 95% CI 1.08 to 4.03, P=0.030). However, missense mutation failed to have a statistically favourable effect on VHL-related survival (HR 0.63, 95% CI 0.38 to 1.03, P=0.067), even though patients with missense mutation still showed a 6 years longer median survival (online supplementary figure b, c and table 2).

We also explored the influence of first presenting symptom on overall survival. Patients in HB (including CHB and RA) group demonstrated a poorer survival than those in abdominal group (62 vs 69 years, P=0.009) (figure 2D). If patients were affected by HB as the first lesion, the risk of death would increase to two times than others affected by abdominal tumours (RCC, PHEO, PCL or PNET) (HR 2.11, 95% CI 1.89 to 3.75, P=0.011). But the effect was not significant in the multivariate Cox regression model (HR 1.54, 95% CI 0.79 to 2.97, P=0.204) (table 2). Moreover, the tendency did not change when we only considered VHL-related death (online supplementary figure d and table 2).

Risk factors for CHB-specific survival in patients with VHL disease

As CHB was the most common cause of death in patients with VHL disease, we next analysed the risk factors for CHB-specific survival. Patients with HB as the first symptom had an obvious worse CHB-specific survival than those with abdominal tumours (P<0.001) (figure 3A). Actually, only two patients died from CHB if their first presenting symptoms were abdominal tumours. In line with overall survival, CHB-specific survival was also correlated to onset age and family history (figure 3B,C). Cox regression analysis showed first presenting symptom, onset age and family history were independent factors for survival when we only considered CHB-related death (HR 8.84, 95% CI 2.04 to 38.37, P=0.004 for first symptom; HR 2.82, 95% CI 1.40 to 5.66, P=0.004 for onset age; HR 2.55, 95% CI 1.08 to 6.00, P=0.032 for family history) (table 3).

Table 3

Univariate and multivariate analyses for risk factors of central nervous system haemangioblastoma-specific survival

Figure 3

The effects of different factors on central nervous system haemangioblastoma (CHB)-specific survival of patients with von Hippel-Lindau disease. Kaplan-Meier curve and log-rank test were used to compare the differences between patient groups by first presenting symptom (A), onset age (B), family history (C) and mutation type (D).


In this study, we assessed the life expectancy of Chinese patients with VHL disease and explored the risk factors for survival of VHL disease. The median survival of patients with VHL disease in China was 62 years, which was still lower than that of the general population in 2015 (76 years).24 VHL disease has been characterised as being with a poor life expectancy in all hereditary cancer syndromes.18 A previous study of an unselected VHL cohort reported that the median survival was only 49 years (n=152).12 However, it just included the affected patients who met the clinical criteria, which could have missed some asymptomatic patients and caused bias. Recently, in line with our results, a national study in Denmark (n=143) showed that the life expectancy improved to 67 years for men and 60 years for women. And they demonstrated that VHL survival was highly dependent on a person’s birth year; the younger the patient is, the longer is the survival.21 In contrast, our study showed that birth year was not an independent risk factor for overall survival (HR 0.99, 95% CI 0.97 to 1.02, P=0.565). This may be partly because we solely recruited patients born after 1930 with accurate clinical data. Another more important reason may be the genetic anticipation in Chinese VHL families; patients in the next generations suffered severer disease than those in the first generation, which counteracted the effect of medical development.25 Therefore, more work needs to be done to improve survival for Chinese patients with VHL disease.

We also observed that female patients had a 7-year longer survival than male patients, although it was not significant. This can be explained by the higher CHB burden and faster tumour growth in men compared with women, considering that CHB was the main cause of death.17 However, Binderup and his team found that female patients had a mortality hazard that were twice as high as male patients, and the effect was more evident for missense mutation carriers (HR 4.75, P=0.004). They attributed the result to hormonal effects on tumorigenesis, but the mechanisms were not clear.21 The inconsistency between the two cohorts is unknown. Future large prospective multicentre studies are needed to confirm the effects of sex on VHL survival.

Type 1 patients showed a lower life expectancy than type 2 patients in this study. The higher risk for CHB and RCC in type 1 VHL disease may explain the result.26 Even so, it is too early to draw the conclusion. First, the two groups were not age-matched. Second, we defined the subtype according to the status of PHEO until patients’ death or the end of follow-up. But patients can move from type 1 to type 2 along with age.26 Actually, of all the type 1 patients (n=290), 40.7% (n=118) individuals were younger than the mean onset age of PHEO (33.7±13.1 years), which could amplify the difference of survival in two subtypes.

We present the first evaluation of the effect of phenotype on VHL survival. Although earlier-onset age was regarded as severer VHL phenotype, no evidence before this study had shown that onset age was correlated to survival. Our results showed that onset age was an independent risk factor for overall survival and VHL-related survival. Our previous study found that early-onset age was correlated to short blood leucocyte telomere length, which reflected the status of genomic instability.27 According to the ‘second-hit’ model of VHL disease,28 patients with an unstable genome will confer a higher age-related risk for loss of function of the wild-type VHL allele, leading to new tumour development. Thus, early-onset patients may suffer from higher tumour burden during their lifetime, which will result in poor survival. On the other hand, we found that patients with HB (RA or CHB) as the first presenting symptom had a lower life expectancy than those with abdominal tumours. Although being with HB as the first presenting symptom is not an independent factor for overall survival, it is a strong and independent risk factor for CHB-specific survival. We combined RA and CHB as one group not only because RA is essentially an HB of CNS considering that retina is usually recognised as a part of CNS, but also there exists no difference in the risk of overall death and CHB-specific death between RA group and CHB group in Cox regression analyses (online supplementary table 3). These results suggest that more active surveillance and treatment approaches are suggested to clinicians for patients with early-onset age and those with HB as the first symptom.

Supplementary Material

Supplementary Table 3

Overall survival of patients with VHL disease is also influenced by genotype and family history. Patients with missense mutation were characterised with a better overall survival, but it was not significant for VHL-related and CHB-specific death. Consistent with our results, another study also found some uncertain results. There existed a significantly lower death rate in missense mutation carriers compared with truncating mutation carriers (HR 0.26, P=0.018), but the effect only existed in male patients. The death rate of female patients showed no difference between genotypes (HR 1.28, P=0.32).21 This can be partly explained by the heavier tumour burden and higher risks of RCC and CHB in truncating mutation carriers,17 29 30 but the effect of genotype needs further confirmation. Patients with a positive family history should have longer life expectancies due to more active surveillance plan in the offspring, but our study draws the opposite conclusion. Genetic anticipation observed in Chinese VHL families may play a role in the unexpected result; patients with VHL disease suffer severer phenotypes in successive generations. This can to some extent explain the poorer survival of patients with a positive family history. However, the number of patients carrying probable de novo mutations (n=70) was small in the present study. Future researches with a bigger size of patient number are needed to confirm the conclusion.

The main cause of death in Chinese patients with VHL disease is CHB, which is consistent with most previous studies. In our study, 67.7% of patients died of CHB, surprisingly higher than the other cohorts. Several studies conducted 20 years age reported that CHB accounted for 33.3%–47.7% of death in patients with VHL disease.15 16 19 The big difference may be due to birth cohort of the patients included. In this study, only 59.1% of patients born before 1950 died of CHB, but the proportion increased to 69.4% and 78.6% for 1950–1970 cohort and after 1970 cohort, respectively (online supplementary table 4). In a recent European study, the overall CHB-specific death proportion is 51%. However, in line with our results, the proportion of CHB-specific death increased from 42% in the 1901–1955 birth cohort to 83% in the 1956–2010 birth cohort.21 This change can be explained by the effect of medical development on the two types of tumours. Clinically, metastasis is the main reason for RCC-related death in patients with VHL disease.31 Thus, the death rate of RCC has decreased along with time for the reason that more and more active surveillance plan is carried out, leading to the diagnosis of RCC before metastasis. As to CHB, many of the deaths occurred during surgery or due to postoperative sequelae.21 Although progress in imaging and neurosurgery has resulted in better prognosis of HBs, the decrease of CHB-specific death risk is not equal to RCC. So, apart from active surveillance, pharmacological and surgical improvement is still urgent for VHL-associated tumours.

Supplementary Material

Supplementary Table 4

The present study is the largest VHL survival analysis to date all over the world. We find that the life expectancy of Chinese patients with VHL disease is 62 years, but it is still lower than the general population. Earlier-onset age, positive family history and truncating mutation type are independent risk factors for poor overall survival and VHL-related survival. Moreover, patients with HB as the first presenting symptom should be screened and treated actively for their high risk of death from CHB.

Supplementary file 1


The authors thank Bu Ding-Fang, Medical Experiment Center, Peking University First Hospital, for his technical assistance, and Yan Ting-Ting, Division of Gastroenterology and Hepatology, Renji Hospital, for her dealing with the graphs.



  • Contributors KG and LC were responsible for the concept and design of the study. J-YW, S-HP and TL dealt with the clinical data. X-HN, S-JL, B-AH and J-YY did the statistical work. WP-Y, P-JW, B-WZ, J-CZ and N-NQ drafted the manuscript. XP, J-FZ and K-FM dealt with the figures and tables. All authors revised the manuscript.

  • Funding This work was supported by the National Natural Science Foundation of China (grant number: 81572506), Beijing Municipal Science and Technology Commission (grant number Z151100003915126) and Beijing Municipal Commission of Health and Family Planning (grant number 2016-2-4074).

  • Competing interests None declared.

  • Patient consent Obtained.

  • Ethics approval This project was approved by the Medical Ethics Committee of Peking University First Hospital (Beijing, China).

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.