The risk of non-melanoma skin cancer in HIV-infected patients: new data and meta-analysis

Abstract

The role of HIV/AIDS in non-melanoma skin cancer (NMSC) is not well defined. We sought to update the evidence of the association between HIV/AIDS and risk of NMSC by gender and antiretroviral therapy (ART). We searched MEDLINE and EMBASE on 29 February 2014. Standardised incidence ratios with corresponding 95% confidence intervals were extracted and combined using generic inverse variance methods assuming a random effects model. Six studies including 78,794 patients with HIV/AIDS fulfilled the inclusion criteria. Analysis of all studies showed that HIV/AIDS was associated with an increased risk of NMSC (standardised incidence ratio 2.76; 95% confidence interval 2.55–2.98). The standardised incidence ratios of NMSC were 3.63 (1.08–12.22) for men and 2.18 (1.24–3.83) for women with HIV/AIDS, respectively. In analysis stratified by ART, we found that individuals receiving ART had lower risk of developing NMSC than individuals who had not received ART (standardised incidence ratio, 95% confidence interval; 1.95 [1.10–3.47] versus 2.11 [1.44-3.12]). HIV/AIDS is associated with an increased risk of NMSC in both male and female patients. The use of ART appears to be beneficial in protecting against the development of NMSC.

Keywords

HIV AIDS skin cancer basal cell carcinoma squamous cell carcinoma highly active antiretroviral therapy epidemiology viral disease human immunodeficiency virus antiviral

Introduction

Non-melanoma skin cancer (NMSC), the most commonly diagnosed cancer in the United States, is among the most costly of all cancers to treat for the Medicare population.^1,2 NMSC consists primarily of basal cell carcinoma (BCC) and squamous cell carcinoma (SCC). BCC is the most common type of NMSC, accounting for up to 80% of cases, with an incidence ranging from as low as 47.4 in women from Scotland to 1541.0 in men from Australia per 100,000 individuals. SCC is the second most common type of NMSC: its incidence per 100,000 individuals ranges from 5.3 in women from Germany to 772.0 in men from Australia.³

HIV and cancer registry linkage studies have provided enough evidence for the increased risks of certain cancer types among persons with HIV/AIDS.^4,5 There has been an increase in the incidence of non-AIDS defining cancers (NADCs) in the HIV-infected population and NADCs now account for the majority of cancers found in persons living with HIV.^6–9 NMSC is the most frequent cancer observed in solid organ transplant recipients.¹⁰ Considering the well-established elevated risk of NMSC in the immunocompromised populations, HIV-infected individuals may be also vulnerable to NMSC.^4,11,12

A previous meta-analysis conducted in 2007 examined cancer incidence in population-based cohort studies of people with HIV/AIDS. Since the meta-analysis was published, a variety of relevant studies on such association have been published.^13–16 However, the association between HIV/AIDS and NMSC remains unclear. To our knowledge, no meta-analysis has examined the association in men and women patients separately. In addition, the relationship between antiretroviral therapy and NMSC is incompletely understood and necessitate further study.

Herein, we conducted an updated meta-analysis to assess the incidence of NMSC among HIV-infected persons. Furthermore, we also evaluated whether the association varied by gender, and examined whether there was evidence that any risk from HIV infection was modified by the use of antiretroviral therapy (ART).

Material and methods

Eligibility criteria

We followed published guidelines for meta-analyses of observational studies.¹⁷ To be eligible for inclusion, studies had to meet the following criteria: (1) they were cohort studies of people with HIV/AIDS; (2) they included mainly adults and (3) they collected data on incident cancer through cancer registers. Studies that accepted other cancer diagnoses without confirming that these were notified to a cancer registry were not included to ensure an unbiased comparison of cohort cancer incidence rates to the general population with standardised incidence ratios (SIRs).⁴

Data sources and searches

A computerised literature search was performed in Medline (from 1 January 1966) and EMBASE (from 1 January 1974), through 29 February 2014, by two independent investigators. We searched the relevant studies with the following text words and/or medical subject headings: ‘HIV’, ‘AIDS’, ‘cancer’, ‘non-melanoma skin cancers’, ‘basal cell carcinoma’, ‘squamous cell carcinoma’ and ‘cohort’. Two authors independently searched through the titles and abstracts to identify relevant papers for inclusion; disagreements were resolved by mutual discussion and no disagreements required adjudication. We also searched the reference lists and conducted a citation search of included studies. There were no articles in languages other than English which, based on the abstract review, met the criteria. For studies that had more than one report, the most recent report of incidence of NMSCs was used where there was a possibility of overlapping data. Figure 1 depicts the process of study selection.

Figure 1.

Flow chart on the articles selection process.

Data extraction

Two investigators independently assessed and extracted the data into a standardised data extraction from each publication. Disagreements were resolved by a third author. We extracted data on demographic and clinical characteristics of the population (age, gender, use of ART, et al.) and the risk estimates with corresponding 95% CIs. The risk estimates were influenced by different confounders in these studies. We assessed whether the effect estimate was adjusted for the effects of potential confounders: smoking, body mass index, use of intravenous street drugs, alcohol consumption, socio-economic status (such as educational attainment, income), CD4 count, viral load or use of ART. From each study, we extracted the risk estimate that was adjusted for the greatest number of potential confounders. For example, the adjusted risk ratio reported in the Silverberg study¹⁶ also include smoking, body mass index and income in addition to age and gender.

Statistical analysis

We included in this meta-analysis studies reporting relative risks (RRs) or SIRs. The variance of the log SIR from each study was calculated by converting the 95% CI to its natural logarithm by taking the width of the CI and dividing by 3.92. Summary relative risk estimates with corresponding 95% CIs were derived using the method of DerSimonian and Laird with the assumptions of a random-effects model, which considers both within-study and between-study variations. When gender-specific estimates were available, we first analysed together (as SIR estimates for NMSC) and then separately (as SIR estimates for NMSC of different gender group).

In assessing heterogeneity among studies, we used the Cochran Q test and I² statistics. These were used to test whether the differences obtained between studies were due to chance. For the Q statistic, a p value of less than 0.10 was used as an indication of the presence of heterogeneity; for I², a value >50% was considered a measure of severe heterogeneity. Sensitivity analysis was performed by excluding each study individually to assess its influence on the overall result of the meta-analysis. We combined studies in subgroups where participants did not use ART or where ART was used and assessed whether use of ART modified the association between HIV infection and incidence of NMSC.

Publication bias was evaluated using a funnel plot of a trial's effect size against the SE. Because funnel plots have several limitations and represent only an informal approach to detect publication bias, we further carried out formal testing using the test proposed by the Begg's adjusted rank correlation test and by the Egger's regression test.^18,19 All statistical analyses were performed using STATA, version 11.0 (STATA, College Station, TX, USA). A two-tailed p value of less than 0.05 was considered to be statistically significant.

Results

Six relevant studies were retrieved, comprising a total of 78,794 patients with HIV or AIDS. Information on country, period of follow-up, cohort entry, number of people with HIV or AIDS and number of cancers is presented in Table 1. One study was based in the United States of America,¹⁶ one study in Scotland,²⁰ one study in England,²¹ one study in Switzerland¹⁴ and two studies in Italy.^13,15 Three studies investigated the incidence of NMSC in HIV-infected or HIV-positive individuals,^15,16,20 two studies assessed the risk of NMSC in people with HIV or AIDS^14,21 and one reported the pattern of cancer risk in persons with AIDS.¹³

Table 1.

Design features of cohort studies of cancer incidence in people with HIV/AIDS.

1st Author, year	Country	Follow-up	Cohort entry	Number of people with HIV/AIDS	Number of cancers
Allardice, 2003²⁰	Scotland	1981–96	HIV registration	2574 HIV-infected individuals	138 AIDS-defining, 24 non-AIDS
Newnham, 2005²¹	England	1985–2001	HIV registration	33,190 people with HIV or AIDS	1580 AIDS-defining, 442 non-AIDS
Dal Maso, 2009¹³	Italy	1986–2005	AIDS registration	21,951 with AIDS	162 AIDS-defining, 383 non-AIDS
Franceschi, 2010¹⁴	Switzerland	1985–2007	HIV or AIDS diagnosis	9429 people with HIV or AIDS	602 AIDS-defining, 256 non-AIDS
Albini, 2013¹⁵	Italy	1999–2009	HIV registration	5090 HIV-infected patients	138 non-AIDS non-virus-related
Silverberg, 2013¹⁶	United States	1996–2008	HIV registration	6560 HIV-positive subjects	386 BCCs, 136 SCCs (67 had both)

BCC: basal cell carcinoma; SCC: squamous cell carcinoma.

As shown in Figure 2, the summary SIR was 2.76 (95% CI 2.55–2.98) in a random-effects model for HIV-infected people. There was significant heterogeneity among these studies (Q = 291.58, p = 0.000, I² = 98.3%). In a sensitivity analysis in which we removed one study at a time and analysed the rest, the SIRs ranged from 4.90 (95% CI, 4.26–5.63; Q = 194.48; p = 0.000; I² = 97.9%) after excluding the study by Silverberg et al.¹⁶ (the study which carried the most weight) to 2.76 (95% CI, 2.55-2.98; Q = 291.58; p = 0.000; I² = 98.6%) after excluding the study by Allardice et al.²⁰ (the study which carried the least weight). Of note, the SIRs for NMSC reported by the Newnham study²¹ were markedly higher than the other studies included in this analysis. The summary SIR was 2.18 (95% CI, 2.01–2.37; Q = 8.12; p = 0.087; I² = 50.8%) with the exclusion of the Newnham study as a sensitivity analysis. Our analysis confirmed the stability of the positive association between HIV/AIDS and NMSC.

Figure 2.

Risk of non-melanoma skin cancer among HIV-infected persons.

Four studies provided results on incidence of NMSC by gender. As shown in Figure 3, HIV/AIDS was associated with an increased risk of NMSC in both male (summary SIR, 3.63; 95% CI, 1.08–12.22) and female (summary SIR, 2.18; 95% CI, 1.24–3.83) patients. Moreover, a significantly stronger positive association was observed in men than in women.

Figure 3.

Comparison of risk of non-melanoma skin cancer by gender.

The temporal sequence between HIV/AIDS and NMSC has not always been clear. Two studies provided results on cancer incidence between the pre- and post-ART periods.^4,5 In addition, one study investigated whether prior ART was associated with the incidence of NMSC.¹ Combining these studies according to ART periods, we found that individuals receiving ART had lower risk of developing NMSC than individuals who had not received ART (summary SIR, 95% CI; 1.95 [1.10–3.47] versus 2.11 [1.44–3.12]) (Figure 4).

Figure 4.

Comparison of risk of non-melanoma skin cancer by antiretroviral therapy.

There was no funnel plot asymmetry for the association between HIV/AIDS and risk of NMSC. p Values for Begg's adjusted rank correlation test was 1.000 and the Egger's regression asymmetry test was 0.508, suggesting a low probability of publication bias (Figure 5).

Figure 5.

Funnel plot of cohort studies evaluating the association between HIV/AIDS and risk of non-melanoma skin cancer.

Discussion

The substantial improvement in survival after HIV infection has led to increasing public health and clinical interest in long-term morbidities in this population.⁶ People infected with HIV and those with AIDS continue to develop many types of cancers more often than people in the general population.⁴ Ten to 15 years is the average time from infection with HIV to depletion of CD4 lymphocytes to a level below 200 cells/µL or the development of opportunistic infection.^4–6

NADCs are now the most frequent cause of cutaneous malignancies among HIV-infected persons.³ The rising incidence and morbidity of NMSC has generated great interest in unravelling of their etiology. The incidence and risk factors associated with NMSC among HIV-infected persons are less defined. Although most NMSCs are easily cured, many become locally invasive and destructive.^2,3 NMSC does not cause substantial social stigma in its early stages because it generally takes a long time to develop. NMSCs rarely metastasise, but when they do, prognosis is poor.^22,23

Grulich et al.⁴ conducted a meta-analysis of four studies in 2007 and showed that immune deficiency was responsible for the increased risk of cancer. Moreover, infection-related cancer will probably become an increasingly important complication of long-term HIV infection. However, the association in different gender groups is worthy of investigation but has not been looked at. Silverberg et al.¹⁶ demonstrated that HIV-positive subjects had a two-fold higher incidence rate of NMSCs compared with HIV-negative subjects. However, there was no difference in the BCC incidence rate by HIV status among women. Although HIV-positive women had an increased SCC incidence rate compared with HIV-negative women, the rate ratio was also not statistically significant.¹⁶

In this meta-analysis, we pooled currently available international studies and found that individuals with HIV/AIDS have an increased risk of NMSC. Moreover, further analysis indicated that the positive association was observed in both male and female patients. In the analyses stratified by gender, the 95% CI for women in the study by Newnham²¹ appears markedly different than that reported in the original study (1.34–38.77 vs. 0.9–26.1 in the original article). As a matter of fact, this was due to the statistical analyses of the 95% CIs. Additionally, the point estimate and 95% CIs for the Silverberg paper¹⁶ appear to reflect the results for men who have sex with men rather than all men (which was not reported in the original study). Our literature search was extensive; we tested for and found no evidence of publication bias. The strength of the study includes that, on an international scale, there are far more individuals with HIV/AIDS in our study.

People with HIV infection are living longer due to the introduction of antiretroviral therapies. ART prolongs the life of HIV/AIDS patients and preserves their immune function, but does not cure or restore the immune system to its pre-infection state. Given the increasing longevity for HIV-positive individuals during the ART era, the burden of many age-related NADCs, including NMSCs, will only continue to increase.²⁴ To determine the relationship between ART and incidence of NMSC, we summed up data from different studies. Of note, individuals who received ART had lower risk of developing NMSC than individuals who had not received ART. However, caution must be taken as evaluation of SIRs does not actually examine this risk. Five of the six studies included can be used to examine differences by period (pre-HAART or post-HAART eras), but differences by period may not only be a result of ART and could also reflect other temporal effects. The study by Silverberg was the only study to specifically examine if ART use was associated with risk of NMSC in HIV-infected populations and while it was not, its effects of higher CD4 + lymphocyte count and suppressed HIV RNA were both protective against NMSC, suggesting that the results of ART are associated with decreased risk.

However, this study has several limitations which should be recognised. First, cohort studies may not be prone to recall bias but are prone to selection bias, because patients with HIV/AIDS are under increased medical surveillance. This bias may distort the true effects.

Second, great heterogeneity exists in terms of study design, gender and use of ART. Of note, when we repeated the analyses with the exclusion of the Newnham study as a sensitivity analysis, the summary SIR was 2.18 (95% CI, 2.01–2.37; Q = 8.12; p = 0.087; I² = 50.8%). Thus, to some extent, the severe heterogeneity in this meta-analysis was due to the results of the Newnham study, which appeared to be an outlier. However, despite the use of appropriate meta-analytic techniques with random-effect models, we could not fully account for these differences.

Third, all the studies included in the meta-analysis were performed in Western countries, and no study was conducted in the Asian population. The six studies included in this meta-analysis includes information on only five (the Dal Maso study¹³ was a national AIDS-cancer linkage study that included data from the Brescia cohort included in the study by Albini¹⁵) high-income, Western and likely racially similar populations where ART was widely available. Even with these similarities, severe heterogeneity was found. As a matter of fact, the income, culture and race of the Asian population are very different from Western countries, and these factors might affect generalisability into other populations. More epidemiological research is warranted to clarify the association between HIV/AIDS and risk of NMSC in the Asian population.

Fourth, only one study provided the adjusted rate ratio for HIV-positive versus HIV-negative subjects for SCCs and BCCs.¹⁶ Further study is needed to assess the risk of SCC and BCC in HIV-infected individuals.

Fifth, important risk factors for the development of skin cancer, regardless of HIV status, include fair skin, family history and cumulative sun-exposure.^25,26 However, our meta-analysis could not assess whether CD4 count, HIV RNA level or these risk factors could impact the development of NMSC among HIV-infected populations.

Sixth, various studies investigated the incidence of NMSC in different population. For example, two studies assessed the risk of NMSC in people with HIV or AIDS^14,21 and one reported the pattern of cancer risk in persons with AIDS.¹³ Newnham et al.²¹ investigated the incidence of NMSC in HIV-infected people in southeast England. Noteworthy, the SIRs for NMSC varied in men and women with HIV infection, AIDS or HIV infection but not AIDS.²¹ Future analyses would involve stratification of the patients by HIV or AIDS status to explore temporal effects on NMSC risk of HIV infection.

Finally, inherent in any meta-analysis of published data is the possibility of publication bias, that is small studies with null results tend not to be published. Publication bias may have resulted in an overestimate of the relationship between HIV/AIDS and risk of NMSC. However, the results obtained from funnel plot analysis and formal statistical tests did not provide evidence for such bias.

In conclusion, our study suggests that HIV/AIDS appears to be associated with a significantly increased risk of NMSC in both male and female patients. There are no strong reasons to doubt the validity of the association. Moreover, the use of ART appears to be beneficial in protecting against the development of NMSC. Although the absolute risks of NMSC are low among HIV-infected individuals, our results have important clinical and public health significance.

Footnotes

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

References

Rogers

Weinstock

Harris

. Incidence estimate of nonmelanoma skin cancer in the United States, 2006. Arch Dermatol 2010; 146: 283–287.

Housman

Feldman

Williford

. Skin cancer is among the most costly of all cancers to treat for the Medicare population. J Am Acad Dermatol 2003; 48: 425–429.

Madan

Lear

Szeimies

. Non-melanoma skin cancer. Lancet 2010; 375: 673–685.

Grulich

van Leeuwen

Falster

. Incidence of cancers in people with HIV/AIDS compared with immunosuppressed transplant recipients: a meta-analysis. Lancet 2007; 370: 59–67.

Silverberg

Chao

Leyden

. HIV infection, immunodeficiency, viral replication, and the risk of cancer. Cancer Epidemiol Biomarkers Prev 2011; 20: 2551–2559.

Crum-Cianflone

Hullsiek

Marconi

. Trends in the incidence of cancers among HIV-infected persons and the impact of antiretroviral therapy: a 20-year cohort study. AIDS 2009; 23: 41–50.

Pantanowitz

Dezube

. Evolving spectrum and incidence of non-AIDS-defining malignancies. Curr Opin HIV AIDS 2009; 4: 27–34.

Engels

. Non-AIDS-defining malignancies in HIV-infected persons: etiologic puzzles, epidemiologic perils, prevention opportunities. AIDS 2009; 23: 875–885.

Simard

Engels

. Cancer as a cause of death among people with AIDS in the United States. Clin Infect Dis 2010; 51: 957–962.

10.

Specchio

Saraceno

Chimenti

. Management of non-melanoma skin cancer in solid organ transplant recipients. Int J Immunopathol Pharmacol 2014; 27: 21–24.

11.

Spanogle

Kudva

Dierkhising

. Skin cancer after pancreas transplantation. J Am Acad Dermatol 2012; 67: 563–569.

12.

Park

Chang

Won

. Incidence of primary skin cancer after organ transplantation: An 18-year single-center experience in Korea. J Am Acad Dermatol 2014; 70: 465–472.

13.

Dal Maso

Polesel

Serraino

. Pattern of cancer risk in persons with AIDS in Italy in the HAART era. Br J Cancer 2009; 100: 840–847.

14.

Franceschi

Lise

Clifford

. Changing patterns of cancer incidence in the early- and late-HAART periods: the Swiss HIV Cohort Study. Br J Cancer 2010; 103: 416–422.

15.

Albini

Calabresi

Gotti

. Burden of non-AIDS-defining and non-virus-related cancers among HIV-infected patients in the combined antiretroviral therapy era. AIDS Res Hum Retroviruses 2013; 29: 1097–1104.

16.

Silverberg

Leyden

Warton

. HIV infection status, immunodeficiency, and the incidence of non-melanoma skin cancer. J Natl Cancer Inst 2013; 105: 350–360.

17.

Stroup

Berlin

Morton

. Meta-analysis of observational studies in epidemiology: a proposal for reporting. Meta-analysis Of Observational Studies in Epidemiology (MOOSE) group. JAMA 2000; 283: 2008–2012.

18.

Begg

Mazumdar

. Operating characteristics of a rank correlation test for publication bias. Biometrics 1994; 50: 1088–1101.

19.

Egger

Davey

Schneider

. Bias in meta-analysis detected by a simple, graphical test. BMJ 1997; 315: 629–634.

20.

Allardice

Hole

Brewster

. Incidence of malignant neoplasms among HIV-infected persons in Scotland. Br J Cancer 2003; 89: 505–507.

21.

Newnham

Harris

Evans

. The risk of cancer in HIV-infected people in southeast England: a cohort study. Br J Cancer 2005; 92: 194–200.

22.

Alam

Ratner

. Cutaneous squamous-cell carcinoma. N Engl J Med 2001; 344: 975–983.

23.

Ganti

Kessinger

. Systemic therapy for disseminated basal cell carcinoma: an uncommon manifestation of a common cancer. Cancer Treat Rev 2011; 37: 440–443.

24.

Shiels

Pfeiffer

Gail

. Cancer burden in the HIV-infected population in the United States. J Natl Cancer Inst 2011; 103: 753–762.

25.

Lobo

Chu

Grekin

. Nonmelanoma skin cancers and infection with the human immunodeficiency virus. Arch Dermatol 1992; 128: 623–627.

26.

Maurer

Christian

Kerschmann

. Cutaneous squamous cell carcinoma in human immunodeficiency virus-infected patients. A study of epidemiologic risk factors, human papillomavirus, and p53 expression. Arch Dermatol 1997; 133: 577–583.