Solid Malignancies Among Patients with Wegener’s Granulomatosis Treated with Etanercept: Long-term Follow-up of a Multicenter Longitudinal Cohort

Abstract

Purpose

An association between therapeutic inhibition of tumor necrosis factor (TNF) and solid malignancies was observed during the Wegener’s Granulomatosis Etanercept Trial (WGET). The present study was conducted to determine the malignancy risk beyond the exposure to study therapy.

Methods

The occurrence and type of solid malignancies were ascertained using a standardized data form. Data collected included vital status, histologic reports, and therapeutic interventions. The SEER database was used to estimate a standardized incidence rate (SIR) for solid malignancies.

Results

The median post-trial follow-up available for 153 patients (85% of the original cohort) was 43 months. Fifty percent of these patients had received etanercept. There were no differences in demographics between etanercept and placebo groups. Thirteen new solid malignancies were detected, 8 in the etanercept and 5 in the placebo group. The risk of solid malignancies in the etanercept group was increased compared to the general population (SIR=3.92; 95% CI 1.69–7.72), but not different from that of the placebo group (SIR=2.89; 95% CI 0.94–6.73, p=0.39). All solid malignancies occurred in patients exposed to cyclophosphamide. The overall duration of disease and a history of malignancy before trial enrollment were associated with the development of malignancy during post-trial follow-up.

Conclusions

The incidence of solid malignancy remained increased during long-term follow-up of the WGET cohort. However, this could not be attributed solely to etanercept exposure during the trial. Anti-TNF therapy with etanercept appears to further increase the risk of malignancy observed in patients with WG treated with cytotoxic therapy and should be avoided in such patients.

Owing to its importance in the mechanisms of inflammation, tumor necrosis factor-alpha (TNF-α) blockers have been widely used for the treatment of immune-mediated, chronic inflammatory diseases. However, the first property ascribed to TNF-α was its ability to induce necrosis of sarcomas and other skin transplanted cancers (1). TNF-α also may play a role in the immunosurveillance against cancer cells by causing cytostasis and cytolysis (2), inducing neoplastic cells to undergo apoptosis (3, 4), and inhibiting tumor-associated angiogenesis (5, 6). There has been a longstanding concern that TNF-α blockers might facilitate the development of neoplasia de novo or the progression of pre-malignant and established malignant lesions by interfering with the normal physiologic effects of TNF-α that control tumor growth.

The relationship between TNF-α blocker use and malignancy has been studied in rheumatoid arthritis (RA), other inflammatory arthropathies, and Crohn’s disease. However, these investigations have been complicated by the increased incidence of malignancies observed in many of those diseases (7–14) and by possible channeling bias (e.g. patients with severe disease that show the highest disease-associated risk of malignancy are more likely to be treated with biologic agents than patients with milder forms of the disease) (15). TNF-α blocker use has been linked to the development of lymphoma, notably hepatosplenic T-cell lymphoma in juvenile inflammatory arthritis and Crohn’s disease that otherwise occurs only rarely (16). In contrast, the association of TNF-α blocker use with solid malignancies remains uncertain (16–21).

For Wegener’s granulomatosis (WG) and microscopic polyangiitis (MPA), the malignancy risk associated with TNF-α blocker use has been even more difficult to estimate because of the low prevalence of these diseases and the paucity of data about TNF-α blocker use in WG and MPA. Furthermore, several reports have suggested an increased risk of both solid and hematologic malignancies in WG and MPA per se, with estimates of global risk ranging from 1.6 – 18 fold compared to the general population (22–24) or patients with other rheumatologic conditions (25, 26). In the Wegener’s Granulomatosis Etanercept Trial (WGET), a placebo-controlled trial in which etanercept or placebo were given in addition to standard therapy for remission induction and maintenance in WG (27), solid malignancies were observed only among patients treated with etanercept who had also been exposed to cyclophosphamide (CYC) with an observed standardized incidence ratio (SIR) of 3.1 [95% CI 1.1–6.8] (28). This finding led to a warning by the manufacturer of etanercept against the concomitant use of etanercept and CYC under any circumstances and in conjunction with any other immununosuppressive agent for the treatment of WG.

To further investigate the relationship between etanercept therapy and malignancy, we conducted a follow-up study to identify and characterize any new cases of solid malignancies that arose in the WGET cohort during the 5 year period after completion of study treatment.

Patients and Methods

The cohort of patients originally enrolled into the WGET forms the basis of this analysis. The design of the WGET, the clinical characteristics of the patients enrolled, and the treatment outcomes have been described in detail previously (27, 29, 30). Briefly, 180 patients who met at least two of the five modified criteria of the American College of Rheumatology for classification of WG were randomized 1:1 to receive either etanercept 25 mg subcutaneously twice weekly or placebo in addition to standard remission induction and maintenance therapy for WG (29, 31, 32). Patients were eligible for enrollment if they had active disease with a BVAS/WG score of ≥3 as a result of newly diagnosed disease or a relapse. Patients who had been diagnosed with a malignancy within 5 years prior to the screening for trial eligibility were excluded from participation, except those with squamous or basal cell carcinomas of the skin or cervical carcinoma in situ who had received curative surgical therapy.

Data were obtained for this study in the context of a long-term follow-up study of the WGET cohort. The occurrence and the type of malignancy after the common closeout (September 2003) were ascertained by investigators in 2007–2008 using a standardized data form. Most patients continue to receive regular follow-up at the original WGET site, and the information was obtained by the physician in direct contact with the WGET participants and/or by reviewing the medical records. If the patient was no longer undergoing regular follow-up at the WGET site, the data form was completed by telephone interview with the patient.

The diagnosis of cancer was confirmed in all cases by reviewing the histopathologic reports. Therapy for WG was recorded for each of these cases, focusing on type and dose of immunosuppressants received before and during the clinical trial, as well as after the trial closeout. Vital status was recorded at the time of last contact, including the date of death if applicable.

Descriptive statistical analyses were performed to analyze the development of malignancies during the trial, after completion of the trial (post-trial follow-up) and taking both periods together (follow-up from trial entry). Exploratory analyses comparing patients who developed solid malignancies versus those who did not were performed using Pearson chi-square analysis or Fisher exact test for categorical variables and Wilcoxon test for continuous variables. Values are expressed as mean (SD), or median (range) unless otherwise specified.

Using the Surveillance Research Program, National Cancer Institute SEER*stat software version 6.3.6, the age- and gender-adjusted incidence rates of solid malignancies were generated for the US population between the years 2000 to 2004 (excluding leukemia, lymphoma, myeloma and non-melanoma skin cancers). This incidence corresponds to the expected rate of solid malignancy for the purpose of this comparison. The incidence rate for malignancies in our cohort was calculated by dividing the number of events by the total patient-years of observation (observed cases). The SIR, [observed/expected cases] × 100) for solid malignancies with the respective 95% CI was calculated.

The SIRs were compared between groups (etanercept versus placebo) using an approximate F test for comparing Poisson variates. With data censored at last follow-up for those without malignancies, the cumulative incidence of malignancies over time following the treatment period was also estimated using the Kaplan-Meier method and compared between groups using the log rank test.

Results

Demographic characteristics of the patients in the cohort

Post-trial follow-up information was available for 153 of the 180 (85%) WGET participants (Figure 1 & Table 1). Their median time of follow-up was 43 months from the common closeout date. Seventy-seven (50.3%) of these patients had been assigned to etanercept in the WGET and 76 (49.7%) to placebo. Gender, age, race, vital status, previous personal or family history of cancer, or current tobacco use did not differ between the etanercept and the placebo groups; only past tobacco use was more common among patients who had received etanercept (29% versus 13%, p=0.015).

Figure 1. Schematic diagram of deaths and solid malignancies (SM) observed during follow-up of the WGET cohort.

(a) Death caused by solid malignancy diagnosed during the trial. (b) Deaths caused by solid malignancy caused after the trial (n=2), leukemia caused after the trial (n=1), myocardial infarction (n=1), indeterminate cause (n=3). (c) Deaths caused by solid malignancy diagnosed after trial (n=2), sepsis (n=2), renal failure (n=1), arrhythmia (n=1), indeterminate cause (n=2).

Table 1.

Demographic characteristics of 153 patients followed after participation in WGET.

Characteristic	Etanercept (E) (n=77)	Placebo (P) (n=76)	Total (n=153)	E versus P p value
Gender, male, n (%)	50 (65%)	44(58%)	94 (61%)	0.37
Age
enrollment, mean (SD)	52 (14)	49 (17)	50 (15)	0.15
last contact, mean (SD)	57 (14)	54 (17)	55 (15)	0.14
Race n (%)
White, non-Hispanic	69 (90%)	72 (95%)	141 (92%)	0.4
Black, non-Hispanic	2 (2.5%)	0 (0%)	2 (1%)	0.5
Hispanic	4 (5%)	2 (2.5%)	6 (4%)	0.4
Other	2 (2.5%)	2 (2.5%)	4 (3%)	0.2
Vital status, alive, (%)	65 (84%)	66 (87 %)	131 (86%)	0.7
Previous history of cancer, n(%)	12 (16%)	5 (7%)	17 (11%)	0.7
Family history of cancer, n(%)	15 (19%)	14 (18%)	29 (19%)	0.9
Tobacco use
Current use, no (%)	1 (1%)	1 (1%)	2 (1%)	0.99
Previous use, no (%)	22 (29%)	10 (13%)	32 (21%)	0.015

Solid malignancies diagnosed since WGET closeout

Thirteen new solid malignancies were diagnosed during the post-trial follow-up period in 13 of the 153 patients (8.5%). The type of cancer and the clinical characteristics of the affected patients are summarized in Table 2. The median time to cancer diagnosis was 52 (31–78) months after trial enrollment and 26 (3–45) months after the common closeout date. The 4 deaths (2 from each treatment group) that occurred during the post-trial follow-up all resulted from the cancer.

Table 2.

Solid malignancies detected in 13 patients after completion of WGET: treatment assignment, disease severity of WG and immunosuppressant drug use.

Case	Arm	Disease extent	Age1/ Gender	Type of cancer	Time to cancer after enrollment/ closeout (m)	Previous cancer/ time before enrollment	CYC USE (grams)			Other drug use (type)			Vital status
							Before trial	During trial	After trial	Before trial	During trial	After trial
1	EG	Limited	29/F	Melanoma	61/27	no	24	20	no		MTX	MTX AZA	Alive
2		Limited	36/M	Squamous cell (tonsillar) carcinoma	52/14	no	183	no	no		MTX	MTX AZA	Alive
3		Limited	71/M	Squamous (transitional) cell, metastatic	56/39	Bladder Ca., 8 years	264	no	no		MTX	MTX	Dead
4		Limited	70/M	Prostate cancer	31/4	No	no	no	yes		MTX	MTX AZA	Alive
5		Severe	56/M	Prostate cancer	50/36	No	no	50	yes		AZA	MTX	Alive
6		Severe	57/M	Melanoma	62/30	No	84	7	no		AZA	AZA	Alive
7		Severe	66/F	Cholangiocarcinoma	38/21	Breast ca, 12 years	48	7	no	AZA	MTX	MTX	Dead
8		Severe	63/M	Small bowel	78/45	Squamous skin ca, >5 years	105	no	no	AZA	MTX	MTX AZA	Alive
9	PG	Limited	62/M	Bladder carcinoma	66/34	No	144	no	no	AZA Cyclo	MTX	MTX	Alive
10		Severe	42/M	Renal cell carcinoma	49/16	No	1	24	no		AZA		Dead
11		Severe	57/M	Colon cancer	35/3	No	2.5	11.9	no		MTX AZA		Dead
12		Severe	61/M	Prostate cancer	34/5	Melanoma, 15 years	64	21	no				Alive
13		Severe	71/F	Cholangiocarcinoma	59/26	No	40	11	no		MTX	MTX AZA	Alive

¹Age at enrollment in WGET.

EG: Etanercept group, PG: Placebo group

CYC: Cyclophosphamide, MTX: Methotrexate, AZA: Azathioprine, Cyclo: Cyclosporine

All of the solid malignancies diagnosed after during the pos-trial period occurred in patients who had been exposed to cyclophosphamide (CYC) before, during, or after the trial (n=138) (Table 2). Twelve of the 13 patients had received at least one additional immunosuppressive agent (AZA, MTX or cyclosporine). The median cumulative dose of CYC used prior to enrollment and during the trial was 56 (1–264) and 16 (7–50) grams, respectively. For the post-trial period, we were able to ascertain the approximate duration of CYC exposure for the individual patients but could not quantify the cumulative doses.

Risk factors for the development of solid malignancies after WGET closeout

We compared the characteristics of the 13 patients who developed solid malignancies after the trial closeout date to those of the 140 patients who did not develop a solid malignancy (Table 3). There were no differences between the two groups in the assigned treatment, age, gender, or extent of disease at trial enrollment. However, patients who developed solid malignancies were more likely to be enrolled into the trial with a disease relapse (85% versus 54%, p=0.04). They also had longer disease duration. Among the group with solid malignancy, the mean time was longer between onset of symptoms and trial enrollment compared with the non-malignancy group (6.4 ± 3.9 years versus 3.3 ± 4.8 years, p=0.001). The same was true for the mean time between diagnosis and enrollment (5.1 ± 4.4 versus 2 ± 3.8 years, p=0.0006). Similar differences were found when the times between symptom onset and diagnosis to end of follow-up were analyzed for both groups (data not shown). Importantly, patients who developed a solid malignancy after trial closeout also had a higher frequency of malignancy before entering the trial (31% versus 9%, p=0.03).

Table 3.

Baseline characteristics of patients who did and those who did not develop solid malignancies after WGET^* trial closeout (September 2003).

	Solid malignancy N=13	No solid malignancy N=140	P value **
Treatment assignment			0.40
Etanercept, no (%)	8 (10)	69 (90)
Control, no (%)	5 (7)	71 (93)
Age at enrollment, mean ± SD, years	57 (13)	50 (16)	0.07
Male/female, no (%)	10 (77)/3 (23)	84 (60)/56 (40)	0.37
Limited/severe disease, no (%)	5 (38)/8 (62)	42 (30)/98 (70)	0.54
Age at onset of vasculitic symptoms, years, mean (SD)	51 (13)	46 (16)	0.26
Relapsing disease, no (%)	11 (85)	75 (54)	0.04
Duration since onset symptoms, years, mean (SD)	6.4 (3.9)	3.3 (4.8)	0.001
Duration since diagnosis, years, years, mean (SD)	5.1 (4.4)	2 (3.8)	0.0006
History of cancer before trial, no (%)	4 (31)	12(9)	0.03
Family history of cancer, no (%)	0 (0)	29 (21)	0.13
Current tobacco use, no. (%)	0 (0)	5 (4)	1.0
Prior treatment wt immunosuppressive drugs, no (%)	11(85)	80(57)	0.08
Ever treated for WG	13 (100)	125 (89)	0.37
CYC treatment, no (%)
Ever used †, daily or intermittent	13 (100)	125 (89)	0.4
During WGET	8 (62)	113 (81)	0.13
MTX treatment, n (%)
Ever used †, oral or SC or IM	9 (69)	82 (59)	0.56
During WGET	11 (85)	110 (79)	1.0
AZA treatment, n (%)
Ever used†	7 (54)	49 (35)	0.23
During WGET	4 (31)	45 (32)	1.0

^*WGET = Wegener’s Granulomatosis Etanercept Trial; SC = subcutaneous; IM = intramuscular.

^**By chi-square or Fisher’s exact test for categorical data, and Wilcoxon’s rank sum test for continuous data.

^†Including before, during or after trial periods.

^##Previous history of cancer was as follows: in the no solid malignancy group, skin basal cell carcinoma (4 patients), colon (3 patients), breast cancer (1), lymphoma (1), pre-cancerous or cancerous skin lesion (1 patient), prostate cancer (1 patient), rectal-skin (1 patient); in the solid malignancy group, bladder (1 patient), breast carcinoma (1 patient), melanoma and basal cell (1 patient) and skin (1 patient).

Relationship between treatment assignment and risk of solid malignancy during WGET follow-up

Table 4 shows the observed and expected number of patients with solid malignancy and the calculated SIR for the etanercept and placebo groups. The analysis is presented for three periods: the trial period (from enrollment until the common closeout in September 2003), the post-trial period (from common closeout until last visit or death) representing the focus of the present study, and the overall period (trial and post-trial periods combined). During the trial, the frequency of solid malignancies was higher in the etanercept group than in the placebo group (7% versus 0%, p = 0.01) (28). However, during the post-trial period the frequency of solid malignancies was not different between the etanercept and placebo groups (10% versus 7%, p=0.39), although the frequency was numerically higher in the etanercept group. When the two periods were combined, the overall frequency of solid malignancies from time of trial enrollment remained higher for the etanercept (18%) compared to the placebo group (7%) (p=0.03).

Table 4.

Frequency of observed and expected solid malignancy events and SIR.

Follow-up period (duration*)	SM etanercept group (E)				SM placebo group (P)				E vs P
	Observed (N)	Expected (N)	SIR	95% CI	Observed (N)	Expected (N)	SIR	95% CI	p†	p††
Trial (30 m)	6	1.58	3.80	1.39 – 8.26	0	1.02	0	0 – 3.63	0.01	0.03
Post-trial (43 m)	8	2.04	3.92	1.69–7.72	5	1.73	2.89	0.94 – 6.73	0.39	0.60
From trial entry (64 m)	14	3.73	3.76	2.05 – 6.31	5	2.92	1.71	0.56 – 3.99	0.03	0.12

^*Duration: median, in months

^†Comparison of frequencies of malignancies, Fisher exact test

^††Comparison of the SIRs, F-test

SM= solid malignancies

N=number of events

SIR=Standardized incidence ratio.

To account for variable age, sex, and duration of follow-up, SIRs were calculated for each treatment group and period (Table 4). For the trial period, the SIR for solid malignancies in the etanercept group was 3.8 [95% CI 1.39–8.26] compared to 0 [95% CI 0–3.63] in the placebo group. These values differ numerically from those previously reported (SIR in etanercept group: 3.12 [95% CI 1.15–6.8]) (28) because the reference population was extended for the present analysis to include the last year with available data in the SEER. For the post-trial period, the SIR for solid malignancies of the etanercept group was not different 3.92 [95% CI 1.69–7.72] from that of the placebo group 2.89 [95% CI 0.94–6.73] (p=0.597). For the combined period, the SIR was 3.76 [95% CI 2.05–6.31] for the etanercept group compared to 1.71 [95% CI 0.56–3.99] for the placebo group (p=0.117). A time-to-event analysis comparing treatment groups confirmed that the occurrence of malignancies post trial (Table 2) did not differ between treatment arms (P=0.44).

We also evaluated whether exposure to other TNF-blockers had an effect on the malignancy rate. Sixteen patients (8 in each WGET treatment group) received infliximab after discontinuation of experimental therapy. One of these patients, assigned to the placebo group, developed a solid malignancy that was previously reported [27]. We reanalyzed the SIRs in two ways: i) by excluding all 16 patients exposed to infliximab from analysis, and ii) by reclassifying the infliximab exposed patients from the placebo group to the etanercept group. Neither analysis resulted in a meaningful change to the SIRs or a significant difference in the malignancy rates between TNF-blocker exposed versus non-exposed patients for any of the observation periods (data not shown).

Discussion

This long-term follow-up study of the WGET cohort shows that in comparison to the general population the increased risk of solid malignancies observed in the etanercept group during the trial persists during post-trial follow-up. However, in contrast to the trial period, solid malignancies were also observed in the placebo group during post-trial follow-up. Furthermore, there was no difference in SIR between the etanercept and placebo groups. Other factors were recognized in association with the development of solid malignancies in the WGET cohort, particularly the duration of disease and a history of malignancy prior to trial enrollment. This suggests that after discontinuation of etanercept the malignancy risk conferred by this agent in WG reverts back to the increased baseline risk inherent to the disease and its treatment. As the duration of disease is linked to duration of therapy, and all solid malignancies that occurred in this cohort developed after CYC exposure, our results further highlight the need for safer treatment regimens, especially ones that reduce or eliminate exposure to CYC.

During the time period of WGET all solid malignancies occurred in the etanercept group (28). In contrast, new malignancies observed during the post-trial follow-up were diagnosed in both treatment arms (Table 4). Based on the SEER database, the risk of solid malignancies remained significantly increased for patients in the etanercept group compared with the general population. However, the relative risk of the placebo group compared to the general population was of similar magnitude. Thus, with longer time from exposure to etanercept, other risk factors become relatively more important.

The SIR of solid malignancies in the etanercept group for the follow-up from trial entry (trial and post-trial period) of our study was 3.76 [95% CI 2.05–6.31], an estimated risk higher than reported for patients with AAV without exposure to TNF-blockers by Westman (standardized morbidity ratio, SMR=1.6 [95% CI 0.9–2.7]) (23), Knight (SIR=2.0 [95% CI 1.7–2.5]) (24), and Faurschou (SIR=2.1 [95% CI 1.5–2.7]) (33). The risk reported in these three studies is similar to the SIR of the placebo group of the WGET cohort (1.71 [95% CI 0.56–3.99]). These studies are comparable to our analysis because they also used healthy populations as reference and risk estimates based on incidence. In contrast, our estimate of risk was lower than those reported by Tatsis (OR=18 [95% CI 2.3–140]) (25) or Pankhurst (RR 6.0 [95% 3.7–9.7]) (26). The latter 12 two studies are not as easily comparable to our analysis because their reference populations consisted of patients with autoimmune diseases (systemic lupus erythematous and RA, respectively) rather than healthy populations, and their estimates were based on the frequency of solid malignancies in the studied groups rather than incidence. In the context of this literature, the observed SIRs of our study suggest that etanercept boosts the already increased background risk of malignancies in patients with WG for the duration of exposure to etanercept, but not much beyond the exposure.

There is ample epidemiologic evidence and mechanistic experimental support linking chronic inflammation to the development of cancer (34). For autoimmune disease in general, it has been speculated that both immune-dysregulation inherent to the underlying disease and the reduction of immune-surveillance caused by immunosuppressive therapy promote malignancy (35, 36). An association between duration of RA and risk of malignancy (lymphoma) has long been suspected, and evidence linking the malignancy risk to disease activity and severity of RA has emerged (11, 37–40).

There is currently no clear evidence that vasculitis per se confers an increased risk for malignancy. For patients with giant cell arteritis, which is not treated with CYC, no increased malignancy risk was found (41). In WG the development of urothelial malignancies has been clearly linked to therapy with CYC, and one recent epidemiologic study has found that the overall cancer risk in WG is linked to the cumulative CYC exposure (33, 42). Of note, at least 10 of the 13 patients who developed solid malignancies after trial close out had received more than 36 g of CYC (Table 2), a cumulative dose identified as critical for the development of malignancies in WG (42).

In our cohort duration of disease and history of previous malignancy emerged as risk factors for the development of solid malignancy. The greater number of different cytotoxic drugs used, and cumulative CYC exposure in the group of patients with solid malignancies, did not reach statistical significance when compared to those who did not develop malignancy (Table 3). However, these factors cannot be separated from disease duration.

The risk of cancer induction by etanercept has been studied extensively in chronic arthropathies, in which the drug has been used widely. An association with hematological malignancies has been suggested, but the estimation of risk is difficult as these autoimmune diseases by themselves have an increased risk of malignancies (especially lymphomas) (12, 43, 44). In contrast, to date, no etanercept-associated risk of solid malignancies has been found in other autoimmune diseases (45, 46). However, in contrast to WG, most patients with RA or other athropathies are usually not treated with CYC.

The malignancies observed during the WGET cohort include a variety of histopathologic types and organs of origin. The fact that 3 (etanercept, 2; placebo, 1) of the 19 solid malignancies diagnosed after enrollment in WGET were cholangiocarcinomas is an unexpected finding owing to the rarity of this neoplasm in the general population.

Our study has several important strengths. The patient cohort consisted of well-characterized patients with a single disease, WG. The exclusion of patients who had a solid malignancy within 5 years prior to enrollment as well as the temporal relationship between the diagnosis of WG and the development of cancer make it unlikely that the vasculitis was a paraneoplastic phenomenon in any of the patients (47). The demographic and clinical characteristics of the patients were well balanced between the etanercept and placebo groups. Finally, the clinicians taking care of the study participants were aware of the possible malignancy risk conferred by randomization to etanercept (28) and of the increased malignancy risk among patients with WG. Consequently, these patients were followed closely, making it unlikely that any solid malignancies were missed in this cohort.

Our study also has limitations. First, follow-up information could only be obtained for 85% of the original cohort, possibly introducing some bias. However, the balance was maintained in the number of patients originally assigned to etanercept versus placebo. The high rate of follow-up likely preserved the balance of risk factors between the two groups (27). Second, the information obtained during post-trial follow-up is less complete than that obtained under the strict trial protocol, but it remains unlikely that any cancers were missed unless they were asymptomatic. Third, the small number of observed solid malignancies precluded a multivariate analysis of risk factors contributing to malignancy development.

In conclusion, the long-term post-trial follow-up of the WGET cohort confirms that patients with WG are at increased risk for solid malignancies. The increased risk compared to the general population observed in the etanercept group during the trial remained unchanged during post-trial follow-up. However, in contrast to the trial period, this increased risk observed during post-trial follow-up can no longer be attributed to an effect of etanercept owing to the lack of a significant difference in the frequency of malignancies between the active treatment and placebo groups. Rather, the increased risk for solid malignancy during post-trial follow-up of this cohort seems to be related to disease duration and history of previous malignancy. Because of the potential interaction between anti-TNF therapy and high CYC exposure in the development of malignancy and the lack of demonstrable efficacy in WG, it is reasonable to avoid the use of etanercept in patients with WG, particularly those with a history of chronic, relapsing disease, CYC exposure, or a previous history of malignancy.