Portal de Periódicos da CAPES

Incidence and Risk Factors Associated with a Second Squamous Cell Carcinoma or Basal Cell Carcinoma in Psoralen + Ultraviolet A Light-treated Psoriasis Patients

2002; Elsevier BV; Volume: 118; Issue: 6 Linguagem: Inglês

10.1046/j.1523-1747.2002.01769.x

1523-1747

Kenneth A. Katz, Isabelle Marcil, Robert S. Stern,

Skin Protection and Aging

Psoralen + ultraviolet A-treated psoriasis patients are at increased risk for squamous cell carcinomas and basal cell carcinomas; however, the incidence and risk factors associated with second squamous cell carcinomas and basal cell carcinomas in this population are not well qualified. Incidence and risk factors for second squamous cell carcinomas and basal cell carcinomas were studied in a cohort of 1380 psoralen + ultraviolet A-treated psoriasis patients prospectively followed for over 20 y; 264 had a squamous cell carcinoma and 258 a basal cell carcinoma after beginning psoralen + ultraviolet A therapy. After a first squamous cell carcinoma, the risk of a second squamous cell carcinoma was 26% at 1 y, 62% at 5 y, and 75% at 10 y. Risk increased with high psoralen + ultraviolet A exposure prior to the first squamous cell carcinoma (hazard ratio 3.32, 95% confidence interval 1.53, 7.18). Higher rates of post-first squamous cell carcinoma psoralen + ultraviolet A treatment also were associated with greater risk (hazard ratio 1.56 for every additional 10 treatments per year for patients with low pre-first squamous cell carcinoma psoralen + ultraviolet A exposure, 95% confidence interval 1.35, 1.81). Patients exposed to high levels of tar and/or ultraviolet B before a first squamous cell carcinoma were also at higher risk (hazard ratio 1.72, 95% confidence interval 1.14–2.60). Risk of a second basal cell carcinoma was 21% at 1 y, 49% at 5 y, and 61% at 10 y. There was some evidence that high exposure to psoralen + ultraviolet A before a first basal cell carcinoma was associated with increased risk of second basal cell carcinoma (hazard ratio 1.45, 95% confidence interval 0.97–2.17). Higher post-first tumor psoralen + ultraviolet A treatment rates also increased risk (hazard ratio 1.24 for every additional 10 treatments per year, 95% confidence interval 1.06–1.47). Psoralen + ultraviolet A-treated psoriasis patients appear to have a greatly increased incidence of second squamous cell carcinoma compared with the general population. Patients who develop a squamous cell carcinoma after starting psoralen + ultraviolet A therapy should be closely monitored for a subsequent squamous cell carcinoma. Psoralen + ultraviolet A-treated psoriasis patients are at increased risk for squamous cell carcinomas and basal cell carcinomas; however, the incidence and risk factors associated with second squamous cell carcinomas and basal cell carcinomas in this population are not well qualified. Incidence and risk factors for second squamous cell carcinomas and basal cell carcinomas were studied in a cohort of 1380 psoralen + ultraviolet A-treated psoriasis patients prospectively followed for over 20 y; 264 had a squamous cell carcinoma and 258 a basal cell carcinoma after beginning psoralen + ultraviolet A therapy. After a first squamous cell carcinoma, the risk of a second squamous cell carcinoma was 26% at 1 y, 62% at 5 y, and 75% at 10 y. Risk increased with high psoralen + ultraviolet A exposure prior to the first squamous cell carcinoma (hazard ratio 3.32, 95% confidence interval 1.53, 7.18). Higher rates of post-first squamous cell carcinoma psoralen + ultraviolet A treatment also were associated with greater risk (hazard ratio 1.56 for every additional 10 treatments per year for patients with low pre-first squamous cell carcinoma psoralen + ultraviolet A exposure, 95% confidence interval 1.35, 1.81). Patients exposed to high levels of tar and/or ultraviolet B before a first squamous cell carcinoma were also at higher risk (hazard ratio 1.72, 95% confidence interval 1.14–2.60). Risk of a second basal cell carcinoma was 21% at 1 y, 49% at 5 y, and 61% at 10 y. There was some evidence that high exposure to psoralen + ultraviolet A before a first basal cell carcinoma was associated with increased risk of second basal cell carcinoma (hazard ratio 1.45, 95% confidence interval 0.97–2.17). Higher post-first tumor psoralen + ultraviolet A treatment rates also increased risk (hazard ratio 1.24 for every additional 10 treatments per year, 95% confidence interval 1.06–1.47). Psoralen + ultraviolet A-treated psoriasis patients appear to have a greatly increased incidence of second squamous cell carcinoma compared with the general population. Patients who develop a squamous cell carcinoma after starting psoralen + ultraviolet A therapy should be closely monitored for a subsequent squamous cell carcinoma. basal cell carcinoma hazard ratio nonmelanoma skin cancer squamous cell carcinoma Since demonstrated in 1974 to be highly effective for psoriasis (Parrish et al., 1974Parrish J.A. Fitzpatrick T.B. Tanenbaum L. Pathak M.A. Photochemotherapy of psoriasis with oral methoxsalen and longwave ultraviolet light.N Engl J Med. 1974; 291: 1207-1211Crossref PubMed Scopus (1160) Google Scholar), orally administered psoralens and long-wave ultraviolet radiation (UVA) (PUVA) have been widely used to treat psoriasis. Non-melanoma skin cancers (NMSC) increasingly account for morbidity and health-care costs as well as mortality (Preston and Stern, 1992Preston D.S. Stern R.S. Nonmelanoma cancers of the skin.N Engl J Med. 1992; 327: 1649-1662Crossref PubMed Scopus (434) Google Scholar). An estimated 1.3 million NMSC [approximately 20% squamous cell carcinomas (SCC) and 80% basal cell carcinomas (BCC)] will be diagnosed in the United States in 2001, making them the most common cancers in that country (Miller and Weinstock, 1994Miller D.L. Weinstock M.A. Nonmelanoma skin cancer in the United States: incidence.J Am Acad Dermatol. 1994; 30: 774-778Abstract Full Text PDF PubMed Scopus (890) Google Scholar;Alam and Ratner, 2001Alam M. Ratner D. Cutaneous squamous-cell carcinoma.N Engl J Med. 2001; 344: 975-983Crossref PubMed Scopus (901) Google Scholar). Psoriasis patients not exposed to carcinogenic therapies do not appear to be at substantially increased risk for SCC or BCC compared with the general population (Stern et al., 1982aStern R.S. Zierler S. Parrish J.A. Methotrexate used for psoriasis and the risk of noncutaneous or cutaneous malignancy.Cancer. 1982; 50: 869-872Crossref PubMed Scopus (67) Google Scholar;Alderson and Clarke, 1983Alderson M.R. Clarke J.A. Cancer incidence in patients with psoriasis.Br J Cancer. 1983; 47: 857-859Crossref PubMed Scopus (24) Google Scholar;Hogan et al., 1989Hogan D.J. To T. Gran L. Wong D. Lane P.R. Risk factors for basal cell carcinoma.Int J Dermatol. 1989; 28: 591-594Crossref PubMed Scopus (79) Google Scholar;Lindelof et al., 1990Lindelof B. Eklund G. Liden S. Stern R.S. The prevalence of malignant tumors in patients with psoriasis.J Am Acad Dermatol. 1990; 22: 1056-1060Abstract Full Text PDF PubMed Scopus (57) Google Scholar); however, compared with the general population, PUVA-treated psoriasis patients have an increased risk of cutaneous cancers, including NMSC, particularly SCC and, to a lesser extent, BCC (Stern et al., 1979Stern R.S. Thibodeau L.A. Kleinerman R.A. Parrish J.A. Fitzpatrick T.B. Risk of cutaneous carcinoma in patients treated with oral methoxsalen photochemotherapy for psoriasis.N Engl J Med. 1979; 300: 809-813Crossref PubMed Scopus (474) Google Scholar,Stern et al., 1984Stern R.S. Laird N. Melski J. Parrish J.A. Fitzpatrick T.B. Bleich H.L. Cutaneous squamous-cell carcinoma in patients treated with PUVA.N Engl J Med. 1984; 310: 1156-1161Crossref PubMed Scopus (377) Google Scholar;Henseler et al., 1987Henseler T. Christophers E. Honigsmann H. Wolff K. Skin tumors in the European PUVA Study. eight-year follow-up of 1,643 patients treated with PUVA for psoriasis.J Am Acad Dermatol. 1987; 16: 108-116Abstract Full Text PDF PubMed Scopus (119) Google Scholar;Stern and Lange, 1988Stern R.S. Lange R. Non-melanoma skin cancer occurring in patients treated with PUVA five to ten years after first treatment.J Invest Dermatol. 1988; 91: 120-124Abstract Full Text PDF PubMed Google Scholar;Forman et al., 1989Forman A.B. Roenigk H.H. Caro W.A. Magid M.L. Long-term follow-up of skin cancer in the PUVA-48 Cooperative Study.Arch Dermatol. 1989; 125: 515-519Crossref PubMed Scopus (78) Google Scholar;Bruynzeel et al., 1991Bruynzeel I. Bergman W. Hartevelt H.M. Kenter C.C.A. Van de Velde E.A. Schothorst A.A. Suurmond D. "High single-dose" European PUVA regimen also causes an excess of non-melanoma skin cancer.Br J Dermatol. 1991; 124: 49-55Crossref PubMed Scopus (90) Google Scholar;Lindelof et al., 1991Lindelof B. Sigurgeirsson B. Tegner E. et al.PUVA and cancer: a large-scale epidemiological study.Lancet. 1991; 338: 91-93Abstract PubMed Scopus (198) Google Scholar;Chuang et al., 1992Chuang T.Y. Heinrich L.A. Schultz M.D. Reizner G.T. Kumm R.C. Cripps D.J. PUVA and skin cancer. a historical cohort study on 492 patients.J Am Acad Dermatol. 1992; 26: 173-177Abstract Full Text PDF PubMed Scopus (93) Google Scholar;Stern and Laird, 1994Stern R.S. Laird N. The carcinogenic risk of treatments for severe psoriasis.Cancer. 1994; 73: 2759-2764Crossref PubMed Scopus (341) Google Scholar). In addition to PUVA, factors reported to increase risk of NMSC among PUVA-treated psoriasis patients include lighter skin type (Forman et al., 1989Forman A.B. Roenigk H.H. Caro W.A. Magid M.L. Long-term follow-up of skin cancer in the PUVA-48 Cooperative Study.Arch Dermatol. 1989; 125: 515-519Crossref PubMed Scopus (78) Google Scholar;Stern and Laird, 1994Stern R.S. Laird N. The carcinogenic risk of treatments for severe psoriasis.Cancer. 1994; 73: 2759-2764Crossref PubMed Scopus (341) Google Scholar), tar, and ultraviolet (UV) B treatments (Henseler et al., 1987Henseler T. Christophers E. Honigsmann H. Wolff K. Skin tumors in the European PUVA Study. eight-year follow-up of 1,643 patients treated with PUVA for psoriasis.J Am Acad Dermatol. 1987; 16: 108-116Abstract Full Text PDF PubMed Scopus (119) Google Scholar;Maier et al., 1996Maier H. Schemper M. Ortel B. Binder M. Tanew A. Honigsmann H. Skin tumors in photochemotherapy for psoriasis. a single-center follow-up of 496 patients.Dermatology. 1996; 193: 185-191Crossref PubMed Scopus (54) Google Scholar), X-ray therapy (Henseler et al., 1987Henseler T. Christophers E. Honigsmann H. Wolff K. Skin tumors in the European PUVA Study. eight-year follow-up of 1,643 patients treated with PUVA for psoriasis.J Am Acad Dermatol. 1987; 16: 108-116Abstract Full Text PDF PubMed Scopus (119) Google Scholar), and arsenic therapy (Honigsmann et al., 1980Honigsmann H. Wolff K. Gschnait F. Brenner W. Jaschke E. Keratoses and nonmelanoma skin tumors in long-term photochemotherapy (PUVA).J Am Acad Dermatol. 1980; 3: 406-414Abstract Full Text PDF PubMed Scopus (73) Google Scholar;Stern et al., 1984Stern R.S. Laird N. Melski J. Parrish J.A. Fitzpatrick T.B. Bleich H.L. Cutaneous squamous-cell carcinoma in patients treated with PUVA.N Engl J Med. 1984; 310: 1156-1161Crossref PubMed Scopus (377) Google Scholar;Maier et al., 1996Maier H. Schemper M. Ortel B. Binder M. Tanew A. Honigsmann H. Skin tumors in photochemotherapy for psoriasis. a single-center follow-up of 496 patients.Dermatology. 1996; 193: 185-191Crossref PubMed Scopus (54) Google Scholar). Methotrexate use was demonstrated to increase risk of NMSC in one report (Henseler et al., 1987Henseler T. Christophers E. Honigsmann H. Wolff K. Skin tumors in the European PUVA Study. eight-year follow-up of 1,643 patients treated with PUVA for psoriasis.J Am Acad Dermatol. 1987; 16: 108-116Abstract Full Text PDF PubMed Scopus (119) Google Scholar) but not in an earlier study of the cohort analyzed in this paper (Stern et al., 1982bStern R.S. Zierler S. Parrish J.A. Psoriasis and the risk of cancer.J Invest Dermatol. 1982; 78: 147-149Crossref PubMed Scopus (41) Google Scholar,Stern et al., 1984Stern R.S. Laird N. Melski J. Parrish J.A. Fitzpatrick T.B. Bleich H.L. Cutaneous squamous-cell carcinoma in patients treated with PUVA.N Engl J Med. 1984; 310: 1156-1161Crossref PubMed Scopus (377) Google Scholar). Ionizing radiation therapy increased risk of NMSC in the cohort analyzed in this study in an early (Stern et al., 1979Stern R.S. Thibodeau L.A. Kleinerman R.A. Parrish J.A. Fitzpatrick T.B. Risk of cutaneous carcinoma in patients treated with oral methoxsalen photochemotherapy for psoriasis.N Engl J Med. 1979; 300: 809-813Crossref PubMed Scopus (474) Google Scholar) but not a later analysis (Stern and Laird, 1994Stern R.S. Laird N. The carcinogenic risk of treatments for severe psoriasis.Cancer. 1994; 73: 2759-2764Crossref PubMed Scopus (341) Google Scholar). Finally, a recent report found cyclosporine use to be associated with an increased NMSC incidence in this cohort (Marcil and Stern, 2001Marcil I. Stern R.S. Squamous-cell cancer of the skin in patients given PUVA and ciclosporin: nested cohort crossover study.Lancet. 2001; 358: 1042-1045Abstract Full Text Full Text PDF PubMed Scopus (196) Google Scholar). Incidence of second NMSC in the general population has been studied (Epstein, 1973Epstein E. Value of follow-up after treatment of basal cell carcinoma.Arch Dermatol. 1973; 108: 798-800Crossref PubMed Scopus (66) Google Scholar;Bergstresser and Halprin, 1975Bergstresser P.R. Halprin K.M. Multiple sequential skin cancers: the risk of skin cancer in patients with previous skin cancer.Arch Dermatol. 1975; 111: 995-996Crossref PubMed Scopus (26) Google Scholar;Robinson, 1987Robinson J.K. Risk of developing another basal cell carcinoma: a 5-year prospective study.Cancer. 1987; 60: 118-120Crossref PubMed Scopus (141) Google Scholar;Schreiber et al., 1990Schreiber M.M. Moon T.E. Fox S.H. Davidson J. The risk of developing subsequent nonmelanoma skin cancers.J Am Acad Dermatol. 1990; 23: 1114-1118Abstract Full Text PDF PubMed Scopus (66) Google Scholar;Frankel et al., 1992Frankel D.H. Hanusa B.H. Zitelli J.A. New primary nonmelanoma skin cancer in patients with a history of squamous cell carcinoma of the skin: implications and recommendations for follow-up.J Am Acad Dermatol. 1992; 26: 720-726Abstract Full Text PDF PubMed Scopus (63) Google Scholar;Marghoob et al., 1993Marghoob A. Kopf A.W. Bart R.S. et al.Risk of another basal cell carcinoma developing after treatment of a basal cell carcinoma.J Am Acad Dermatol. 1993; 28: 22-28Abstract Full Text PDF PubMed Scopus (88) Google Scholar) and recently reviewed and subject to a meta-analysis (Marcil and Stern, 2000Marcil I. Stern R.S. Risk of developing a subsequent nonmelanoma skin cancer in patients with a history of nonmelanoma skin cancer. a critical review of the literature and meta-analysis.Arch Dermatol. 2000; 136: 1524-1530Crossref PubMed Google Scholar). Second NMSC among PUVA-treated psoriasis patients, however, have not been systematically assessed. Utilizing data from a cohort prospectively studied for more than two decades, this analysis quantifies the incidence and risk factors of second SCC and BCC among PUVA-treated psoriasis patients. The PUVA Follow-up Study enrolled 1380 psoriasis patients at 16 centers in the United States in 1975 and 1976. Data for this analysis were collected through July 1, 1999. During each of 18 follow-up cycles since enrollment, patients were telephoned and asked about number of PUVA treatments received as well as about health events, including skin cancer diagnoses. Dates, types, body locations, and numbers of reported skin tumors were confirmed by pathology reports. Patients were asked about: (i) past X-ray and arsenic therapy at the time of enrollment; (ii) tar, UVB, and methotrexate therapy at entry and during each follow-up cycle; and (iii) cyclosporine use during each follow-up cycle. This analysis includes all patients who had at least 6 mo of follow-up after a first postenrollment BCC or SCC. Separate analyses were done for SCC and BCC. Only prospectively ascertained SCC (not including SCC in situ or keratoacanthomas) or BCC occurring after first treatment with PUVA were considered in the analysis. Tumors diagnosed during a 180 d "wash-out" period after a first tumor, including multiple tumors diagnosed on the same day as the index tumor, were not considered second tumors in this analysis. For incidence as well as hazard ratio calculations, time at risk was defined as beginning 6 mo after diagnosis of a first SCC or BCC (i.e., after the "wash-out" period) and ending with the diagnosis of a second SCC or BCC, respectively, or with loss to follow-up. Tumor dates and locations on the body were taken from pathology reports. Only tumors thought to be new primaries, rather than suspected recurrences, were counted as second tumors. Age was the age at development of the first tumor. Skin types were categorized in two groups, I–II or III–VI. Regions of the United States where patients initially received treatment were categorized as northern, middle, or southern, as in prior analyses (Stern et al., 1979Stern R.S. Thibodeau L.A. Kleinerman R.A. Parrish J.A. Fitzpatrick T.B. Risk of cutaneous carcinoma in patients treated with oral methoxsalen photochemotherapy for psoriasis.N Engl J Med. 1979; 300: 809-813Crossref PubMed Scopus (474) Google Scholar). Body locations of first tumors were categorized as more (head or neck) or less (rest of body) sun exposed. The number of PUVA treatments received through the interview cycle during which the first tumor was diagnosed was categorized as low (0–159), medium (160–259), or high (260+). Exposure to X-ray therapy or arsenic therapy reflected patient self-reporting. Cyclosporine exposure was considered positive if a patient had used cyclosporine for at least three consecutive months. High tar and/or UVB exposure corresponded to 45 or more months of tar treatments and/or 300 or more UVB treatments by 1989, provided the first SCC or BCC was diagnosed by 1989. Similarly, high methotrexate exposure reflected more than 3 y of therapy, again provided a first SCC or BCC diagnosis by 1989. Logistic regression models were initially fitted using the raw number of post-first tumor PUVA treatments. Preliminary results that showed, counterintuitively, a protective effect of increasing PUVA treatments were hypothesized to reflect the fact that patients who did not develop second tumors had more time to receive more PUVA treatments. To correct for this, the number of post-first tumor PUVA treatments was divided by the number of years of follow-up, yielding a post-first tumor PUVA treatment rate that was used in all subsequent analyses. The rate was modeled using two variables, allowing simultaneous estimation of effects of a rate of zero, or an increasing rate of, PUVA treatments after the first tumor. The reference group for this variable throughout the analyses is patients with a rate of nearly (but not) zero PUVA treatments per year. For univariate analyses, all variables were individually entered linearly into Cox proportional hazards models (Cox, 1972Cox D.R. Regression models and life tables.J R Stat Soc. 1972; 34B: 187-220Google Scholar;Armitage and Berry, 1994Armitage P. Berry G. Statistical Methods in Medical Research. 3rd edn. Blackwell Science, Oxford1994Google Scholar;Collett, 1994Collett D. Modelling Survival Data in Medical Research. Chapman & Hall, London1994Crossref Google Scholar). In each analysis, only variables trending toward significance in univariate analyses (p < 0.15) were entered into a multivariate model. Backward-stepwise procedures (Altman, 1991Altman D.G. Practical Statistics for Medical Research. Chapman & Hall, London1991Google Scholar) were used to identify variables significantly related to second tumor risk in multivariate analyses. Interactions between all variables in final models were tested for significance using likelihood ratio tests. Proportional hazards assumptions for each model were tested using score residuals (Collett, 1994Collett D. Modelling Survival Data in Medical Research. Chapman & Hall, London1994Crossref Google Scholar;Stata Reference Manual, 1999Stata Reference Manual, Version 6. Stata Press, College Station, TX1999Google Scholar). Kaplan–Meier and predicted survival functions were generated using standard techniques. All statistical analyses were performed with Stata 6.0 (Stata, College Station, TX). Of 1380 PUVA Follow-up Study patients, 276 had a postenrollment SCC and 265 a postenrollment BCC before July 1, 1999. Of these, 264 with an SCC and 258 with a BCC had at least 6 mo of follow-up and were included in this analysis. Characteristics of these cohorts are shown in Table I. For the SCC cohort, median overall follow-up time was 2.5 y (minimum 0.5, maximum 22.5, 1080.2 total person-years of follow-up). For BCC, it was 3.3 y (minimum 0.5, maximum 21.5, 1326.6 total person-years of follow-up).Table ICharacteristics of psoriasis patients in the PUVA Follow-up Study who developed a SCC or a BCC and had at least 6 mo of follow-upPatients with a first SCC (n = 264)Patients with a first BCC (n = 258)Mean age at first tumor (SD)60.7 11.961.1 (12.5)Men (%)198 (75.0)184 (71.3)Skin type (%) 1–280 (30.3)82 (31.8) 3–6157 (59.5)155 (60.1) Missing27 (10.2)21 (8.1)Region of residence at time of initial survey (%) North139 (52.7)144 (55.8) Middle43 (16.3)47 (18.2) South81 (31.1)67 (26.0) First tumor located on head or neck (%)44 (16.7)67 (26.0)Number of PUVA treatments before first tumor (%) Low (0–159)91 (34.5)117 (45.4) Medium (160–259)74 (28.0)61 (23.6) High (260+)99 (37.5)80 (31.0)Number of PUVA treatments after first tumor Patients with 0 treatments (%)145 (54.9)164 (63.6) If > 0, median number of treatments (minimum, maximum)75 (1, 464)65.5 (1, 485) History of X-ray therapy for psoriasis at time of entry into study80 (30.3)75 (29.1) History of arsenic therapy for psoriasis at time of entry into study10 (3.8)10 (3.9) History of tar treatment (45+ months) and/or UVB treatments (300+) as of January 1, 198984 (31.8)85 (33.0) History of methotrexate therapy (3+ years) as of January 1, 198981 (30.7)75 (29.1) History of cyclosporine therapy (3 or more consecutive months) during follow-up8 (3.0)9 (3.5)Percentage of with a second tumor of same type (95% confidence interval) at: 1 y26.2 (21.3, 32.0)21.1 (16.6, 26.7) 3 y49.7 (43.5, 56.2)40.8 (34.8, 47.3) 5 y62.1 (55.7, 68.7)49.2 (43.8, 55.9) 10 y75.1 (68.4, 81.2)61.4 (54.5, 68.4) 15 y90.8 (81.2, 96.7)69.9 (61.0, 78.3) Open table in a new tab Based on Kaplan–Meier estimates (Figure 1), the percentages with a second SCC at 1, 3, 5, 10, and 15 y were 26.2%, 49.7%, 62.1%, 75.1%, and 90.8%, respectively. For a second BCC the corresponding percentages were 21.1%, 40.8%, 49.2%, 61.4%, and 69.9%, respectively. In univariate analyses for both cohorts (Table II), PUVA exposure prior to a first tumor was significantly related to risk of a second tumor of the same type. The rate of post-first tumor PUVA exposure was also associated with second tumor risk for both cohorts. Among patients who received a minimal (but nonzero) rate of post-first tumor PUVA treatments, risk of a second SCC or BCC increased linearly with the rate; however, for both tumor types, risk was, counterintuitively, higher among patients who received no post-first tumor PUVA treatments than those who received low rates. High tar and/or UVB exposure prior to the first SCC was significantly related to risk of developing a second SCC, and there was some evidence of a similar relationship for BCC. Treatment with methotrexate prior to developing a first tumor was a significant risk factor for a second SCC.Table IIUnivariate analysis of potential risk factors associated with development of a second SCC or a second BCC in patients in the PUVA Follow-up StudyPotential risk factorPatients with a first SCC (n = 276)Patients with a first BCC (n = 265)Hazard ratio95% CIp-valueHazard ratio95% CIp-valueAge at first tumor, for every additional 10 y1.010.89, 1.150.861.130.99, 1.300.08Men, compared with women1.200.84, 1.720.311.220.83, 1.770.31Skin types III–VI, compared with types I–II0.980.70, 1.370.901.430.97, 2.100.07Region of residence, compared with0.70ap-value for region covariate overall.0.02ap-value for region covariate overall. north middle1.190.79, 1.800.540.32, 0.92 south1.000.71, 1.401.120.77, 1.63First tumor not on head or neck1.160.77, 1.740.481.380.91, 2.080.13Pre-first tumor PUVA treatment category, compared with low category (< 160 treatments)< 0.01bp-value for model of pre-first tumor PUVA treatment categories, fitted linearly.0.04bp-value for model of pre-first tumor PUVA treatment categories, fitted linearly. medium (160–259 treatments)1.371.14, 1.651.231.02, 1.50 high (260+ treatments)1.881.30, 2.721.521.03, 2.25Post-first tumor PUVA treatment rate< 0.01cp-value for model of post-first tumor PUVA treatment rate, fitted using two variables (as explained in text), tested jointly for statistical significance.< 0.01cp-value for model of post-first tumor PUVA treatment rate, fitted using two variables (as explained in text), tested jointly for statistical significance. 02.111.52, 2.944.312.63, 7.10 If > 0, for every additional 10 treatments per year1.131.10, 1.171.281.09, 1.51X-ray therapy prior to entering study1.210.87, 1.680.261.170.82, 1.670.40Arsenic therapy prior to entering study0.870.38, 1.960.740.660.27, 1.610.36High tar and/or UVB exposure prior to developing first SCC/BCC1.851.25, 2.74< 0.011.540.98, 2.420.06High methotrexate exposure prior to developing first SCC/BCC1.561.04, 2.400.031.450.91, 2.320.12Cyclosporine use prior to developing first SCC/BCC1.840.68, 4.980.231.640.60, 4.440.33Cyclosporine use after developing first SCC/BCC2.400.88, 6.500.090.930.29, 2.910.90a p-value for region covariate overall.b p-value for model of pre-first tumor PUVA treatment categories, fitted linearly.c p-value for model of post-first tumor PUVA treatment rate, fitted using two variables (as explained in text), tested jointly for statistical significance. Open table in a new tab In multivariate analysis (Table III), high pre-first tumor PUVA exposure remained a significant risk factor in both cohorts. Patients with high pre-first SCC PUVA exposure had a hazard ratio of 3.32 (95% CI 1.53, 7.18) compared with those with low pre-first SCC PUVA exposure, assuming that both had very low (but nonzero) post-first SCC PUVA treatment rates.Table IIIRisk factors associated with a second postenrollment SCC in patients in the PUVA Follow-up Study, from multivariate analysisRisk factorHazard ratio95% CIp-valuePre-first SCC PUVA exposure category, compared with low (< 160 treatments), for patients with a PUVA treatment rate of nearly (but not) zero after the first SCC< 0.01aP-value for pre-first SCC PUVA treatment category, post-first SCC PUVA treatment rate category, and terms for interactions between those two variables, tested jointly. Medium (160–259 treatments)2.231.01, 4.92 High (260+ treatments)3.321.53, 7.18Post-first tumor PUVA treatment rate0 for patients with low pre-first SCC PUVA exposure3.641.75, 7.56 for patients with medium pre-first SCC PUVA exposure2.451.28, 4.69 for patients with high pre-first SCC PUVA exposure2.721.63, 4.55If > 0, for every additional 10 treatments per year for patients with low pre-first SCC PUVA exposure1.561.35, 1.81 for patients with medium pre-first SCC PUVA exposure1.141.07, 1.22 for patients with high pre-first SCC PUVA exposure1.121.06, 1.18Tar/UVB use prior to developing first SCC1.721.14, 2.600.01a P-value for pre-first SCC PUVA treatment category, post-first SCC PUVA treatment rate category, and terms for interactions between those two variables, tested jointly. Open table in a new tab For BCC (Table IV), evidence for an association between pre-first tumor PUVA exposure and risk of a second BCC was of borderline statistical significance (p = 0.07). Nevertheless, because this finding was biologically plausible, it was kept in the multivariate model. For second BCC, patients with high pre-first SCC PUVA exposure had a hazard ratio of 1.45 (95% CI 0.97, 2.17) compared with patients with low pre-first BCC PUVA exposure, regardless of the post-first BCC PUVA treatment rate for either patient.Table IVRisk factors associated with a second postenrollment BCC in patients in the PUVA Follow-up Study, from multivariate analysisRisk factorHazard ratio95% CIp-valuePre-first tumor PUVA treatment category, compared with low category (< 160 treatments)0.07ap-value for model of pre-first tumor PUVA treatment categories, fitted linearly. medium (160–259 treatments)1.200.98, 1.47 high (260+ treatments)1.450.97, 2.17Post-first tumor PUVA treatment rate< 0.01bp-value for model of post-first tumor PUVA treatment rate, fitted using two variables (as explained in text), tested jointly for statistical significance. 04.182.54, 6.88 If > 0, for every additional 10 treatments per year1.241.06, 1.47a p-value for model of pre-first tumor PUVA treatment categories, fitted linearly.b p-value for model of post-first tumor PUVA treatment rate, fitted using two variables (as explained in text), tested jointly for statistical significance. Open table in a new tab Higher rates of post-first tumor PUVA treatments were also significant risk factors for both cohorts. For the SCC cohort only, risk associated with post-first tumor rates depended on pre-first tumor PUVA exposure (p < 0.01 for the interaction). The additional risk associated with every 10 treatment rate increase was highest among patients who had low pre-first tumor PUVA exposure (HR 1.56, 95% CI, 1.35, 1.81) and lowest among patients with high pre-first tumor PUVA exposure (HR 1.12, 95% CI 1.06, 1.18). As in univariate analyses, patients in either cohort who received no PUVA treatments after a first tumor were at higher risk than those who received low rates of post-first tumor PUVA treatment. For second SCC, the