Article Text

Cancer and neurologic degeneration in xeroderma pigmentosum: long term follow-up characterises the role of DNA repair
  1. Porcia T Bradford1,
  2. Alisa M Goldstein1,
  3. Deborah Tamura2,
  4. Sikandar G Khan2,
  5. Takahiro Ueda2,
  6. Jennifer Boyle2,
  7. Kyu-Seon Oh2,
  8. Kyoko Imoto2,
  9. Hiroki Inui2,
  10. Shin-Ichi Moriwaki2,
  11. Steffen Emmert2,
  12. Kristen M Pike3,
  13. Arati Raziuddin3,
  14. Teri M Plona3,
  15. John J DiGiovanna2,4,
  16. Margaret A Tucker1,
  17. Kenneth H Kraemer2
  1. 1Genetic Epidemiology Branch, Division of Cancer Epidemiology & Genetics, National Cancer Institute, Bethesda, Maryland, USA
  2. 2Dermatology Branch, National Cancer Institute, Bethesda, Maryland, USA
  3. 3Laboratory of Molecular Technology, National Cancer Institute at Frederick, Science Applications International Corporation, Frederick, Maryland, USA
  4. 4Dermatology Department, The Warren Alpert Medical School of Brown University, Providence, Rhode Island, USA
  1. Correspondence to Dr Kenneth H Kraemer, DNA Repair Section, Dermatology Branch, Center for Cancer Research, National Cancer Institute, Building 37 Room 4002 MSC 4258, Bethesda, MD 20892-4258, USA; kraemerk{at}


Background The frequency of cancer, neurologic degeneration and mortality in xeroderma pigmentosum (XP) patients with defective DNA repair was determined in a four decade natural history study.

Methods All 106 XP patients admitted to the National Institutes of Health from 1971 to 2009 were evaluated from clinical records and follow-up.

Results In the 65 per cent (n=69) of patients with skin cancer, non-melanoma skin cancer (NMSC) was increased 10 000-fold and melanoma was increased 2000-fold in patients under age 20. The 9 year median age at diagnosis of first non-melanoma skin cancer (NMSC) (n=64) was significantly younger than the 22 year median age at diagnosis of first melanoma (n=38)—a relative age reversal from the general population suggesting different mechanisms of carcinogenesis between NMSC and melanoma. XP patients with pronounced burning on minimal sun exposure (n=65) were less likely to develop skin cancer than those who did not. This may be related to the extreme sun protection they receive from an earlier age, decreasing their total ultraviolet exposure. Progressive neurologic degeneration was present in 24% (n=25) with 16/25 in complementation group XP-D. The most common causes of death were skin cancer (34%, n=10), neurologic degeneration (31%, n=9), and internal cancer (17%, n=5). The median age at death (29 years) in XP patients with neurodegeneration was significantly younger than those XP patients without neurodegeneration (37 years) (p=0.02).

Conclusion This 39 year follow-up study of XP patients indicates a major role of DNA repair genes in the aetiology of skin cancer and neurologic degeneration.

  • Genetic epidemiology
  • DNA repair
  • skin cancer
  • neurologic degeneration
  • xeroderma pigmentosum
  • dermatology
  • neurology
  • cancer: dermatological

Statistics from

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.


Evaluation of outcomes in patients with the neurocutaneous disorder, xeroderma pigmentosum (XP), has been limited by the number of individuals with this rare disorder and the heterogeneity of the clinical manifestations. XP is autosomal recessive with defects in repairing DNA damaged by ultraviolet (UV) radiation leading to a dramatically increased acute skin hypersensitivity to minimal sun exposure in some (figure 1A), but not all (figure 1B), patients and the development of non-melanoma skin cancer (NMSC) (figure 1C) and melanoma at early ages.3–6 Some XP patients also have progressive neurologic degeneration (figure 1D) with some features of premature ageing.4 6–9 XP is heterogeneous, resulting from different defects in the nucleotide excision repair (NER) pathway.7 10–13 Seven XP complementation groups (genes), XP-A through XP-G, are associated with defective NER. The remaining group, XP variant, is deficient in DNA polymerase eta which is involved in translesional DNA synthesis.

Figure 1

Xeroderma pigmentosum (XP) patients in study. (A) Patient XP420BE complementation group XP-D at 9 months of age with severe blistering erythema of the malar area following minimal sun exposure. Note sparing of her forehead and eyes that were protected by a hat. (B) Patient XP358BE (XP-C) at age 2 years did not sunburn easily but developed multiple hyperpigmented macules on her face. A rapidly growing squamous cell carcinoma (SCC) or keratoacanthoma grew on her upper lip and a pre-cancerous lesion appeared on her forehead. (C) Northern African patient XP393BE (XP-C)1 at age 23 years with numerous hyperpigmented macules on his face. Nodular basal cell cancer is present on his left nasal root. Pigmented basal cell cancer is present on his left cheek. His eyes show cornea scarring from unprotected sun exposure. (D) Patient XP19BE (XP-A)2 at age 35 years with neurological degeneration. He has numerous hyperpigmented macules on sun exposed areas of his face and neck. Progressive sensorineural deafness requires use of a hearing aid.

We assessed the effects of defective DNA repair genes on the frequency of skin cancer, neurologic degeneration, and mortality in XP patients examined at the National Institutes of Health (NIH) beginning in 1971 and ending in 2009. In addition, we examined the influence of genetic variations in melanocortin 1-receptor (MC1R), a gene strongly associated with human pigmentation, melanoma, and NMSC in the general population14–17 and in melanoma-prone families,18 on the risk of skin cancer in XP patients.


Data sources

Data were abstracted from medical records at the Clinical Center, NIH; from medical records of outside institutions; from research records at the National Cancer Institute (NCI); from the Social Security Death Index; from previous publications; and from personally related information.


We conducted a retrospective follow-up study of clinically confirmed XP patients initially examined by a co-author (KHK) at the NIH Clinical Center from 1971 to 2009. Patients were classified as having XP, XP/Cockayne syndrome complex (XP/CS), or XP with trichothiodystrophy (XP/TTD) as previously described.4 6 7 9 19 Patients were generally referred by healthcare professionals. Informed consents were obtained under NCI Institutional Review Board approved protocols in effect at enrolment.

End points

Demographic and clinical data

The primary end points assessed were vital status, skin and other cancer occurrence, neurologic degeneration, and cause of death. Follow-up was obtained via telephone conversation with patients, family members, physicians, and review of the Social Security death index ( Clinical data were collected using a standard questionnaire5 completed at the time of the initial visit to NIH, medical records including pathology reports of skin and other cancers, patient photographs, research records, and published reports.1–7 9 20–46 Only cancers validated by pathology report review were included. Variables collected included vital status, age, sex, ethnicity, age at first skin cancer, number and types of skin and other cancers, XP4 6 7 9 43 and non-XP type neurologic abnormalities, and cause of death.

Melanocortin-1 receptor (MC1R) variations

A family based case–control study was conducted on a subset of subjects having appropriate DNA samples to evaluate the relationships between MC1R variants in XP patients compared to unaffected family members. Sequencing of the 951 base pair MC1R coding region was performed at the Laboratory of Molecular Technology, SAIC, Frederick, Maryland, USA, as described previously.18 Variants were detected using Mutation Surveyor (SoftGenetics Inc, Pennsylvania, USA) and SAIC sequence analysis software.

XP complementation group status

XP complementation group assignment was performed using cell fusion, host cell reactivation, western blotting or DNA sequencing.1 6 21–23 25 27–29 31 34–37 39 45–47

Statistical analyses

Cancer risk and mortality

For comparison with general population rates of cancer or mortality, follow-up began at birth, and ended on the date of diagnosis of cancer for the cancer rates, the date the patient was last known to be cancer-free (for cancer outcome), the date of death for the mortality analyses, the date last known to be alive (for mortality), or the end of follow-up (31 December 2009), whichever came first. The observed number of NMSC was compared to the expected number from the Kaiser Permanente skin cancer database48 after adjustment for age, sex, race, and birth cohort. The observed number of melanoma skin cancers was compared with the expected number in the general population (O/E ratio) based on the NCI Surveillance, Epidemiology, and End Results (SEER) cancer database.49

The relative risk (RR) for each cause of death was estimated by calculating the standardised mortality ratio (SMR) and the exact Poisson 95% confidence interval (CI). The expected number of deaths was calculated by applying the US mortality rates (by 5 year age, five calendar year, and sex specific categories) to the appropriate person-time accrued by XP survivors in the cohort.50 Kaplan–Meier estimates were used to compare survival between different groups. PROC LIFETEST of SAS (version 9.1.3) was used to test for differences in survivor function using the Wilcoxon test. Results with p<0.05 were regarded as significant. All statistical tests were two-sided.

MC1R and skin cancer risk

We evaluated each MC1R variant individually comparing 1+ variant to the consensus MC1R sequence. Because many MC1R variants were too rare to examine their individual associations with skin cancer risk, we also used the following MC1R variables in the analyses: carriers of any MC1R variant compared with wild-type MC1R; carriers of multiple (1,2+) variants compared with the consensus sequence; carriers of 1 non-red hair colour (NRHC) variant, 2+ NRHC variants, 1 red hair colour variant (RHC), 2+ RHC variants, or carriers of both RHC and NRHC variants compared with wild-type MC1R.18 We also examined whether MC1R variants influenced the number of skin cancers the patient had as well as whether the number of MC1R variants influenced the age at diagnosis of the patient's first skin cancer. χ2 and Fisher exact tests were used to measure the association between skin cancer risk and RHC variables. The non-parametric Jonckheere–Terpstra test51 was used to test the hypothesis of no differences among the ages at diagnosis of skin cancer according to number or type of MC1R variants against the alternative that the ages at diagnosis decreased as number or type of MC1R variants increased.


Study demographics

All 106 XP patients examined at the NIH Clinical Center from January 1971 through December 2009 were included in this retrospective study (table 1). Of these, 97 were classified as XP, two as XP/CS complex, and seven as XP/TTD; 78 were non-Hispanic whites (NHW). The median age at last observation or death was 26 years. The age at the last NIH visit was used for the three patients who were lost to follow-up. Always/sometimes burning on minimal sun exposure was reported in 63% of the patients (n=65) (table 1, figure 1A and supplemental table 1).

Table 1

Xeroderma pigmentosum patients at the National Institutes of Health, 1971–2009

Complementation group assignments were determined for 100 of the patients1 6 21–23 25 27 29–31 34–37 39 42 43 45 46 and unpublished observations (table 1 and online table 1). XP-C was the most common complementation group (43%, n=46) followed by XP-D (n=30). No XP-F or ERCC-1 patients were examined at NIH although defects in these genes were identified in cell lines from patients with late onset severe neurological degeneration who were not seen at NIH.52 As we reported previously,27 cells from six XP patients could not be assigned to any of the known complementation groups and may thus have defects in presently unidentified genes.

Skin cancer analysis

Sixty-nine patients (65%) had one or more skin cancers and 64 of these patients had NMSC (table 1, figures 2 and 3 and supplemental table 1). Some patients had hundreds of primary skin cancers—for example, we reported that patient XP29BE had 284 basal cell carcinoma (BCC), 12 squamous cell carcinoma (SCC), and 24 melanomas documented histologically.22 45 Melanoma was diagnosed in 38 patients; 33 of these also had NMSC (table 1, figure 2). The NMSC preceded or was diagnosed at the same time as the melanoma in 29 of the 31 patients where age at first NMSC and melanoma was known (figure 3B).

Figure 2

Xeroderma pigmentosum (XP) skin cancer by age at first skin cancer diagnosis and skin cancer type compared to US general population. Upper panel: Proportion of non-melanoma skin cancer (NMSC) patients diagnosed at selected ages. Lower panel: Proportion of melanoma patients diagnosed at selected ages. Individuals with both NMSC and melanoma were used for both analyses. General population data taken from Glass and Hoover.48

Figure 3

Skin cancer and mortality in xeroderma pigmentosum (XP) patients. (A) Probability of the absence of skin cancers for all XP patients (n=106). At age 12 years, 50% of the patients had been diagnosed with non-melanoma skin cancer (NMSC) or melanoma skin cancer (arrows). (B) Scatter plot age of diagnosis of first skin cancer in patients with both NMSC and melanoma. NMSC was diagnosed earlier or at the same time as melanoma in 29/31 patients. (C) Probability of the absence of skin cancer stratified by burning phenotype. Patients who ‘never’ burned on minimal sun exposure (n=38) were significantly more likely to develop NMSC or melanoma skin cancer at an earlier age than those who ‘always or sometimes’ burned (n=65) on minimal sun exposure (p=0.006). (D) Probability of the absence of skin cancer stratified by XP complementation group. Patients in complementation groups XP-A, XP-B, XP-D, and XP-G developed skin cancer at a significantly older age than those in complementation groups XP-C, XP-E, and variant (p=0.009). (E) Kaplan–Meier curve of XP patient survival compared to US general population. 30% of XP patients had died by age 32. The survival of the XP patients was significantly less than the general population (p<0.001). (F) Kaplan–Meier curve of XP patient survival stratified by neurologic phenotype. Patients with neurologic degeneration had poorer survival rates than those without neurologic degeneration (p=0.04).

Among the XP patients, the median age at diagnosis of first NMSC was 9 years (range 1–32 years). The median age at diagnosis of first melanoma (22 years, range 2–47 years) was significantly older than NMSC (p<0.001). The age at diagnosis of XP skin cancer was substantially reduced compared to the general population: NMSC (median age 67 years) and melanoma (median age 55 years) (figure 2A,B). There were 57 XP patients diagnosed with NMSC under the age of 20 years, producing a RR about 10 000 (95% CI 7388 to 12 639) (table 2) times the US population. Similarly, there were 14 XP patients diagnosed with melanoma under the age of 20 years, resulting in a RR more than 2000 (95% CI 1027 to 3154) (table 2) times the US population.

Table 2

Observed/expected rates of skin cancer in xeroderma pigmentosum patients by age

NMSC was present in 60% of the NHW patients (table 1) (median age at diagnosis 9 years, n=47). Similarly 44% of the African/African American (A/AA) patients had NMSC (median age 12 years, n=7) (p=0.8). Two A/AA patients (but no NHW patients) had SCC of the anterior tongue (p=0.04). No skin melanomas were diagnosed in the A/AA patients (p=0.1).

By the age of 13 years, 50% of the patients had developed skin cancer (figure 3A). When stratified by burning phenotype, those XP patients who never burned were more likely to be diagnosed with skin cancer at an earlier age than those who had a history of always or sometimes burning on minimal sun exposure (p=0.006) (figure 3C). This difference was also found for age at first NMSC and age at first melanoma. Many patients in complementation group XP-C, XP-E, and variant never burned but were more likely to have skin cancer than patients in complementation groups XP-A, XP-B, XP-D, and XP-G (p<0.001) (figures 1B,C and 3D, supplemental table 1). Some of the XP-D and XP-A patients had severe blistering burns after a few minutes sun exposure (figure 1A), but this extreme sun sensitivity was not reported for XP-C patients.

Neurological status

Progressive neurologic degeneration was observed in 25 (24%) XP patients. This included loss of intellectual functioning, deterioration of neurologic status, impaired hearing, abnormal speech, areflexia, ataxia, peripheral neuropathy, and loss of ability to walk and talk. Pathologically this is due to a primary neuronal loss reflected as cortical atrophy and dilated ventricles on MRI.4 6 7 9 43 Patients with progressive neurologic degeneration were primarily in complementation group XP-D (n=16) and XP-A (n=6) (supplemental table 1). Two patients had XP/CS complex with skin features of XP and neurological degeneration of CS, including retinal degeneration with defects in the XPB and XPG genes6 9 34 36 37 (table 1). Ten patients had non-XP related neurologic abnormalities; these included XP-C patients with hypoglycinaemia,28 38 with hearing loss and intellectual impairment without loss of coordination30 or with neurological disorders and systemic lupus erythematosus.53

Patient mortality

Table 3 shows causes of death for the 29 XP patients who died. The median age of death was 32 years. XP survival rates were significantly lower than in the general population (p<0.001) (figure 3E). The major causes of death were skin cancer (34%, n=10), neurologic degeneration (31%, n=9), and internal cancer (17%, n=5). Six deaths were secondary to metastatic melanoma (SMR=14, 95% CI 3 to 25), and four were secondary to invasive SCC (SMR=38, 95% CI 1 to 75). The SMR of death from internal cancers was 16.4 (95% CI 4.3 to 28.6). All six internal cancers occurred in patients who also had many skin cancers. There were three central nervous system cancers (SMR=11, 95% CI 1 to 23), and a peripheral nerve cancer (table 3).24 These occurred in XP-C patients who had no neurological abnormalities. We previously reported that patient XP3BE, who died from lung cancer at age 37 years, had smoked cigarettes since age 18 years and also had many skin cancers and metastatic melanoma.5 6 42 Similarly, our patient XP1BE had more than 100 skin cancers including NMSC and melanomas, but died at age 49 years from metastatic endocervical adenocarcinoma of the uterus.6 40 43

Table 3

Xeroderma pigmentosum patient deaths, 1971–2009

XP patients with neurologic degeneration had poorer survival rates than those patients who had no neurologic degeneration. The median age at death (29 years) in XP patients with neurological degeneration was significantly lower than those XP patients without neurological degeneration (37 years) (p=0.02; figure 3F). The median age at death was 32 years in XP patients with defects in complementation groups that tended to be associated with neurologic degeneration (complementation groups XP-A, XP-B, XP-D and XP-G) compared to age 37 years in XP patients with defects in complementation groups less associated with neurologic degeneration (XP-C, XP-E and variant) (p=0.05). Deaths in patients with defects in the XPC gene were predominantly from cancer (metastatic or locally invasive skin cancer or nervous system cancers). In contrast, deaths in patients with defects in the XPD gene were predominantly related to progressive neurologic degeneration despite having large numbers of skin cancers.

MC1R analysis

We included 79 XP patients (cases) and 101 unaffected XP heterozygotes/family members (controls) in analyses of the role of MC1R variants on the risk of skin cancer. As noted in previous studies of the general population,59–61 NHW had the majority of MC1R variants among these XP families (supplemental table 2).

There were no significant differences in the number or types of MC1R variants between cases and controls for any of these comparisons for all subjects or for NHW subjects (data not shown). MC1R variants were not associated with the occurrence of skin cancer in XP patients (all or NHW only), the number of skin cancers (single vs multiple), or the type of skin cancers (NMSC vs melanoma) (data not shown). Contrary to expectation, patients with RHC variants were significantly older at first melanoma diagnosis than those who did not have RHC variants (age=32 years, n=5, vs age=22 years, n=15; p=0.01); however, these data involved only 20 patients with melanoma. Overall this small study of XP patients suggests that defective DNA repair has a greater influence on skin cancer risk than variants of MC1R.


In 1968 James Cleaver reported defective DNA repair in XP.62 At NIH, Jay Robbins then initiated a study to evaluate XP patients; Kenneth Kraemer joined the study in 1971 and has led it for decades. We report an up to 39 year follow-up including 15 patients reported in 19746 plus additional patients through 2009. We found a >10 000-fold increased risk of NMSC and >2000-fold increased risk of melanoma under age 20. These rates, which vary somewhat from our earlier studies that included some of these patients,3 5 are based on longer follow-up and increased numbers, thus contributing substantially more person years at risk. The occurrence of UV type mutations in the tumour suppressor genes p53 in NMSC63 and PTEN in melanomas56 provide molecular evidence of a direct effect of UV exposure in skin cancer in XP patients. Compared to the general population,48 64–67 the XP patients had a 58 year reduction in age at first NMSC, and a 33 year reduction in age at first melanoma. As in the general population, we found that the anatomic site distribution of NMSC in the XP patients was different from that of melanomas.5 These differences suggest differences in mechanisms of carcinogenesis between NMSC and melanoma and emphasises the importance of DNA repair in the protection against NMSC. Indeed, we found that XP-C patients with only a few per cent of normal XPC mRNA resulting from a splice lariat mutation have a lower frequency of NMSC skin cancer than other XPC patients with different splice mutations, leading to undetectable levels of XPC mRNA.46 68

Acute ultraviolet (UVB) exposure of the skin produces sunburn, an inflammatory response with erythema and blistering characterised histologically by a mixed dermal neutrophilic and lymphocytic infiltrate.69 In the general population sunburning is a skin cancer risk factor.70 Surprisingly, 38 of the XP patients reported never burning but these XP patients were more likely to be diagnosed with skin cancer at an earlier age than the 61% who had a history of always/sometimes burning on minimal sun exposure. This may be partly related to early initiation of rigorous sun protection because XP patients often experience severe blistering sunburn on minimal exposure (figure 1A) (primarily in XP-A and XP-D). XP patients in complementation groups XP-A, XP-B, XP-D, and XP-G (those with higher frequency of neurologic disease) were more likely to develop skin cancers at a later age than those patients in complementation groups (XP-C, variant) with no neurologic disease. These patients may also have less mobility. Many XP-C patients did not report burning on minimal sun exposure but tan normally and develop freckle-like pigmentation at an early age followed by skin cancer.30 Similarly, mice with a defect in the XPC gene do not burn on minimal UV exposure.71 Fibroblasts from XP patients with a history of sunburning on minimal exposure (in XP-A and XP-D) were reported to be more sensitive to killing by UV than fibroblasts from XP patients who did not burn easily.20 40 It is thus possible that transcription coupled DNA repair (which is defective in XP-A and XP-D but not XP-C)13 mediated cytokine generation26 44 72 may play a major role in generation of the inflammatory sunburn response.

The role of pigmentation in protection from skin cancer is complex. In agreement with other studies of A/AA patients,73–75 cancers occurred on less pigmented sites, including the anterior tongue, at a greater frequency than in the NHW patients.1 3 5 This sun-exposed site is an extremely unusual location for tongue neoplasms.76 Perhaps the dark skin offers some protection despite defective DNA repair; thus these patients may experience greater sun exposure than lighter skinned patients. The frequency of NMSC in A/AA (3/100 000) is about 100-fold lower than in NHW.77 In contrast, in A/AA XP patients the frequency of NMSC (44%) was not significantly different from that in NHW patients (61%). Thus, a normally functioning DNA repair system may provide greater protection from NMSC than the dark pigmentation present in A/AA skin.

The most common cause of death was skin cancer (metastatic melanoma or invasive SCC). Despite an early age of skin cancer diagnosis, ∼45% live into their 40s and the oldest patient died at age 73 years. As reported by others,63 78 we found an increased mortality rate from central nervous system (CNS) tumours in patients with defects in the XPC gene who did not have XP neurological degeneration. There is some evidence that these tumours may result from exposure to oxidative damage.63 78 Lung cancer was present in two patients who smoked cigarettes, which may be a consequence of the hypersensitivity of XP cells to mutagenic effects of components of cigarette smoke.13

Progressive neurologic degeneration, which may have resulted from primary degeneration of previously normally developed neurons,7–9 was a major cause of death among the 25% of the susceptible patients (in XP-D, XP-A, XP-G, and XP-B). Fibroblasts from XP patients with neurological degeneration were reported to be more sensitive to killing by UV than cells from XP patients without neurological degeneration.20 41 These findings implicate a role for DNA repair in maintenance of the viability of neurons.

Specific variants of the pigmentation related gene, MC1R, are associated with increased UV induced erythema response79 and increased risk of both melanoma and NMSC14–17 in the general population, and in families with mutations in the melanoma susceptibility gene, CDKN2A.18 In contrast, based on limited data, MC1R variants do not appear to dramatically affect the risks of skin cancer in individuals with XP.

Patients in this NIH study may not reflect the general population of XP patients. Many of our patients have been protected from the sun from very early ages, which could influence their subsequent development of skin cancer. This 39 year study characterises the major morbidity and mortality of XP and suggests a major role of DNA repair in the aetiology of skin cancer and neurologic degeneration.


We are indebted to Dr Jay Robbins for initiating these XP studies. We thank Joe Zou and Nathan Appel (Information Management Services, Inc) for assistance with statistical analyses; Caren Nadem, Engin Gozukara, and Tala Shalavi (Kraemer lab) who helped in complementation group analysis; and Scott Coccodrilli, John Elser, Viktoriya Grinberg, Robin Stewart, Hue Vong, and David Sun (Laboratory of Molecular Technology, NCI) for MC1R analysis.


Supplementary materials


  • Previous presentations: This study was presented on 8 May 2010 at the American Dermatoepidemiology Network Symposium during the Annual Meeting of the Society of Investigative Dermatology in Atlanta, Georgia, USA. An abstract was published in the Journal of Investigative Dermatology 2010;130:S61.

  • Funding This study was supported by the Intramural Research Program of the National Institutes of Health, National Cancer Institute, Division of Cancer Epidemiology and Genetics and the Center for Cancer Research.

  • Competing interests None declared.

  • Patient consent Obtained.

  • Ethics approval This study was conducted with the approval of the National Cancer Institute, NIH, Bethesda, MD.

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