Single-center series and systematic review of randomized controlled trials of malignancies in patients with rheumatoid arthritis, psoriatic arthritis, and ankylosing spondylitis receiving anti–tumor necrosis factor α therapy: Is there a need for more comprehensive screening procedures?

Abstract

Objective

To systematically review the occurrence of malignancies among patients with rheumatoid arthritis (RA), psoriatic arthritis (PsA), and ankylosing spondylitis (AS) treated with anti–tumor necrosis factor α (anti-TNFα) therapy in randomized controlled trials (RCTs), and to report a retrospective personal case series evaluating the frequency of malignancies in patients with RA, PsA, and AS requiring anti-TNF therapy selected with more comprehensive cancer screening procedures compared with patients screened according to previously published procedures.

Methods

The primary outcome was the report of frequency of malignancies in RCTs and the latency between the therapy introduction and the occurrence of the neoplasm. A total of 363 consecutive RA, PsA, and AS patients requiring anti-TNF therapy from 2002 to 2006 observed at the Rheumatology Unit in Prato, Italy, underwent extensive cancer screening procedures. An historical controlled group of 73 patients treated between January 1999 and December 2001 underwent the screening procedures accepted for the RCT procedures.

Results

Thirty-six RCTs were included for analysis. Malignancies occurred in 60 (0.75%) of 8,015 patients randomized to the active treatment arm and in 21 (0.52%) of 3,991 patients in the placebo arms (P = 0.15). In the personal retrospective case series, 1 study patient (0.27%) and 3 controls (4.1%) developed cancer over the followup period (P = 0.017). Mean ± SD followup duration was 40.9 ± 16.7 months in study patients and 50.6 ± 18.1 months in controls.

Conclusion

The results of RCTs and our data showing 26% of malignancies occurring within 12 weeks from enrollment suggest the need for a revision of current cancer screening procedures in RCTs and in clinical practice.

INTRODUCTION

Randomized controlled trials (RCTs) have provided consistent evidence on the efficacy of the anti–tumor necrosis factor α (anti-TNFα) drugs infliximab, etanercept, and adalimumab to treat rheumatoid arthritis (RA), psoriatic arthritis (PsA), and ankylosing spondylitis (AS) (1–36). These medications are currently employed in clinical practice, with important benefits in terms of improvement of signs and symptoms as well as of radiographic progression inhibition (37–40).

Because TNFα exerts an important role in host defense and in the pathobiology of cancer through its action on natural killer cells and CD8 lymphocyte–mediated killing of tumor cells, an increase of malignancy occurrence has been considered as a possible adverse event of TNFα blockade (41, 42).

A recent meta-analysis of RCTs of infliximab and adalimumab reported a significantly higher occurrence of solid tumors in patients receiving the active medication compared with placebo arms (43). Other studies did not confirm this finding (44, 45).

Data from the Swedish Register and from one Japanese, one Canadian, and 3 US health care databases (46–51) seem to exclude an increased frequency of all malignancies in patients receiving anti-TNFα agents compared with the general population and with those taking traditional disease-modifying antirheumatic drugs.

The Italian Ministry of Health approved the use of infliximab, etanercept, and adalimumab to treat RA, PsA, and AS during the years 2000–2005. However, we began to employ these medications in October 1999 by compassionate use or in clinical trials (27, 35, 52–54). During the period from October 1999 to December 2001, we recorded the occurrence of cancer in 2 patients with RA and in 1 patient with PsA after a few infusions of infliximab. Due to the precocity of malignancy onset, we considered these events to be unrelated to anti-TNF therapy, but rather related to inadequate cancer screening procedures. Consequently, in January 2002, we decided to adopt a more comprehensive set of exclusion criteria for patients treated in daily clinical practice as well as for those recruited for controlled clinical trials.

The aim of this study was to review the occurrence of malignancies in RCTs discussing the adopted screening procedures. The retrospective evaluation of the occurrence of neoplasms in a cohort of patients with RA, PsA, and AS treated with anti-TNFα drugs during the past 8 years in our center was used to support the data obtained from the review of RCTs.

MATERIALS AND METHODS

Literature review.

Study selection, assessment of eligibility criteria, data extraction, and statistical analysis were performed based on a prespecified protocol. One of the investigators (CN) searched Ovid EMBase, Ovid Medline, and the Ovid Cochrane Library from 1998 to September 2008 using the following terms: rheumatoid arthritis, psoriatic arthritis, ankylosing spondylitis, infliximab, adalimumab, etanercept, safety and efficacy assessment, anti-TNF alfa therapy, randomized clinical trials, multicenter studies, clinical trials phase III, clinical trials phase IV.

To locate unpublished trials, we searched the electronic abstract databases of the annual scientific meetings of the European League Against Rheumatism and the American College of Rheumatology (ACR). Only articles written in English were included, and we only included RCTs that lasted for at least 12 weeks. Open-label randomized trials and open-label extensions of RCTs were excluded.

Assessment of eligibility criteria for inclusion or exclusion and extraction of outcome variables was performed independently by 2 investigators (CN and FC) with an intraobserver agreement kappa statistic of 1.

Retrospective personal case series analysis.

Setting.

The retrospective evaluation of the occurrence of malignancies was conducted by one of the investigators (FC) in the Rheumatology Unit of Prato Hospital, Tuscany, Italy, which is a secondary referral center that serves ∼300,000 people living in the province of Prato and the surrounding industrial areas. Since January 1998, the demographic and clinical data of all new outpatients have been recorded and regularly updated in a computed individual chart. The study was approved by the local Ethical Committee and written informed consent was obtained from all study patients and controls.

Study population.

We retrospectively evaluated the frequency of malignancies in patients with RA, PsA, and AS treated with anti-TNFα agents from January 2002 to December 2006 after the adoption of more comprehensive cancer screening procedures with respect to those used in RCTs.

The patients treated with anti-TNF agents were under the care of the investigators. The decision to begin an anti-TNF agent was at the discretion of the individual physicians and based on current approval standards for these medications. Physicians initiated these agents as part of the usual standard of care.

Control group.

We used an historical control group of patients with RA, PsA, and AS receiving the same medications from 1999 to 2001 who were screened on the basis of a diagnosis of cancer or a positive history of cancer over the previous 10 years.

We reviewed the clinical records of all new, consecutive outpatients meeting the 1987 revised ACR (formerly the American Rheumatism Association) criteria for the classification of RA (55), the modified New York criteria for AS (56), and for those with PsA, defined as an inflammatory arthritis that is usually seronegative for rheumatoid factor associated with psoriasis (57); and who were treated with etanercept 50 mg/week subcutaneously, infliximab 3–5 mg/kg every 8 weeks intravenously, and adalimumab 40 mg every other week subcutaneously during the period October 1999 to December 2006. In this cohort, we also included patients treated with anti-TNF agents during the unblinded phase of RCTs and other open studies that our center has participated in (27, 35, 52–54).

Outcome measure and treatment schedules.

The primary outcome measure for this study was the frequency of cancer occurrence after starting therapy.

In 1995, we adopted a standardized therapeutic approach for patients with RA, PsA, and AS. RA patients were initially given methotrexate (MTX) at a dosage of 10–15 mg/week associated with nonsteroidal antiinflammatory drugs (NSAIDs) and low-dose corticosteroids. Combination therapy with cyclosporin A 3–5 mg/kg/day or sulfasalazine 2 gm/day was scheduled in nonresponders. After 1999, MTX-resistant patients received infliximab or etanercept and, after its approval, adalimumab.

PsA patients presenting with mono-oligoarthritis were initially treated with NSAIDs, short-term low-dose corticosteroids, and local infiltrative therapy when indicated. Before 2003, cyclosporin A at a dosage of 3–5 mg/kg/day was added to MTX in nonresponders, with the exception of 16 patients enrolled in a clinical trial (52). Corticosteroids were permitted in the case of resistance to therapy. After approval of infliximab and etanercept use for PsA in 2003 and adalimumab in 2005, nonresponders to MTX and to combined therapy with MTX and cyclosporin A received one of the 3 TNFα blockers. PsA patients presenting with polyarthritis were scheduled to start MTX and NSAIDs at diagnosis. Nonresponders were treated with the same schedule as described above.

AS patients initially received NSAIDs and low doses of corticosteroids for at least 3–4 months. In nonresponders, sulfasalazine 2 gm/day was added. During the years 2001 to 2005, infliximab and successively etanercept and adalimumab were approved to treat AS, and we employed these drugs according to the ASsessment in Ankylosing Spondylitis (International Working Group) recommendations (58).

Patients resistant or intolerant to one anti-TNF drug were switched to another. In the case of resistant patients, before switching, dose escalation of infliximab from 3 mg/kg to 5 mg/kg with infusion interval shortening to 6 weeks, and dose escalation of adalimumab to 40 mg/week were applied.

Safety and cancer screening procedures.

Except for those with cancer, before starting anti-TNF therapy all patients underwent the safety screening procedures adopted in RCTs (1–36).

Controls (October 1999 to December 2001) had either a current diagnosis of cancer or a history of cancer during the previous 10 years according to the cancer screening procedures in RCTs.

Study patients (January 2002 to December 2006) underwent a modified set of screening procedures as follows (59): a current diagnosis of malignancy or a positive history of cancer, abdominal ultrasound (US) examination, and hormone replacement therapy (HRT) in postmenopausal women and contraceptive pill cessation before the start of anti-TNF therapy, with no permission to use these medications during the treatment (60). For subjects age ≥50 years, detection of fecal occult blood (3 hemoccult tests on samples from 3 consecutive days) was performed; determination of serum tumor markers, including carcinoembryonic antigen, CA-19-9 antigen, β₂-microglobulin, CA-15-3 antigen, cancer antigen 125 test, and prostate-specific antigen, was performed in men, and these markers were considered abnormal if values were 10% higher than the normal upper limits; and in heavy smokers (≥20 cigarettes/day for ≥20 years), lung computed tomography (CT) beyond chest radiograph was performed for tuberculosis screening.

These procedures were chosen somewhat arbitrarily but in keeping with standard practices for age- and sex-related cancer screening. Current cancer or history of cancer has been used in most RCTs.

Abdominal US was used to detect potential subclinical intraabdominal or retroperitoneal masses or lymphadenopathy (61, 62). The known association of HRT with breast cancer influenced our decision to exclude women who were currently receiving that type of treatment (63). The procedures related to those individuals age >50 years are standard recommendations for cancer screening, with the exception of CT scans of the chest. We believed that this latter screening test was appropriate for smokers in whom a plain chest radiograph might not detect an abnormality.

In the presence of any abnormality of one or more of these items, patients were initially excluded from the treatment and they underwent appropriate clinical investigation. If no malignancy was detected, patients started the therapy and the same investigation was repeated at 12-month intervals. The same screening procedures, with the exception of lung CT, were also repeated every 12 months in patients without any contraindication to treatment at first visit.

Followup.

At every visit, patients had a complete physical examination and laboratory tests, including acute-phase reactants, a complete blood cell count with differential count, renal and liver function tests, and antinuclear antibodies.

Patients were followed by the same rheumatologist, with scheduled visits at baseline and every 4 months. This interval was shortened in the case of urgent clinical problems, and all patients were instructed to call the center in the presence of worsening of arthritis, extraarticular manifestations, or adverse events. The end of the followup period was extended to June 2007.

Adverse events.

At every visit, all patients were monitored for clinical and laboratory evidence of adverse events, defined as mild (transient and easily tolerated), moderate (subject discomfort with interruption of usual activities), or severe (incapacitating or life-threatening).

Statistical analysis.

Descriptive statistics and statistical differences were calculated using Microsoft Excel Software of the Microsoft Office Package for Windows XP, version 2003 (Microsoft, Redmond, WA). Continuous variables were summarized by the mean ± SD, and categorical variables were displayed as numbers and percentages. T-tests for continuous variables and Pearson's chi-square test or Fisher's exact test for categorical variables were used to calculate the differences between the study patients and the controls. P values less than 0.05 were considered significant.

RESULTS

Systematic review results.

We analyzed all published RCTs with at least a 12-week duration of infliximab, etanercept, and adalimumab employed for the treatment of RA, PsA, and AS (1–36). As shown in Table 1, 81 malignancies (0.67%) were recorded in 12,006 patients recruited in 36 RCTs. Malignancies occurred in 60 (0.75%) of 8,015 patients randomized to the active treatment arm and in 21 (0.52%) of 3,991 patients in the placebo arms (P = 0.15). Infliximab was employed in 7 RCTs of RA (1–7), 2 RCTs of PsA (8, 9), and 4 RCTs of AS (10–13), with 3,564 patients recruited. In the study drug arms, 14 malignancies (0.55%) occurred in 2,535 patients, and 4 malignancies (0.39%) occurred in 1,029 patients assigned to placebo (P = 0.719). Etanercept trials were in 9 RCTs of RA (14–21, 36), 2 RCTs of PsA (22, 23), and 5 RCTs of AS (24–28), with a cumulative number of 4,943 patients. Malignancies were observed in 23 (0.72%) of 3,197 patients in the study drug arms and in 11 (0.63%) of 1,746 patients in the placebo arms (P = 0.718). Finally, adalimumab was studied in 5 RCTs of RA (29–33), 1 RCT of PsA (34), and 1 RCT of AS (35), with recruitment of 3,499 patients. Malignancy occurrence was reported in 24 (1.0%) of 2,283 patients receiving the study drug and in 6 (0.49%) of 1,216 placebo-treated controls (P = 0.134).

Table 1. Malignancy occurrence in the study drug and placebo arms of randomized controlled trials on the efficacy and safety of infliximab (IFX), etanercept (ETN), and adalimumab (ADA) in patients with RA, PsA, and AS*

Author, year (ref.)	Disease	Patients enrolled, no.	Active treatment/no.	Control patients/no.	Active treatment arm		Placebo arm		Study duration, weeks
					Malignancy	Interval from enrollment, weeks	Malignancy	Interval from enrollment, weeks
* RA = rheumatoid arthritis; PsA = psoriatic arthritis; AS = ankylosing spondylitis; MTX = methotrexate; NA = not available; IMPACT = Infliximab Multinational Psoriatic Arthritis Controlled Trial; SSZ = sulfasalazine; NOS = not otherwise specified; DMARDs = disease-modifying antirheumatic drugs; MALT = mucosa-associated lymphoid tissue; COMET = Combination of Methotrexate and Etanercept in Active Early Rheumatoid Arthritis Trial. † Patient with a skin cancer history. ‡ Myelodysplastic syndrome at enrollment. § Loss of appetite, weight loss, and night sweats prior to study entry. ¶ Quoted by Bongartz et al (43).
Maini et al, 1998 (1)	RA	101	IFX + MTX/43	Placebo + MTX/14	None	0	None	0	26
			IFX + placebo/44
Maini et al, 1999 (2) and Lipsky et al, 2000 (3)	RA	428	IFX + MTX/340	Placebo + MTX/88	1 B cell lymphoma	26	None	0	54
					1 breast cancer (recurrence)	19
					1 melanoma + 1 squamous cell carcinoma	26
					1 basal cell carcinoma (recurrence)	8
					1 rectal carcinoma	26
St.Clair et al, 2004 (4)	RA	1,049	IFX + MTX/841	Placebo + MTX/208	1 endometrial cancer	3	None	0	54
					1 pancreatic carcinoma with metastases	18
					1 colon cancer	52
					1 acute myeloid leukemia	52
Quinn et al, 2005 (5)	RA	20	IFX/10	10	None	NA	None	0	54
Westhovens et al, 2006 (6)	RA	1,084	IFX + MTX/721	Placebo + MTX/363	1 lung squamous cell carcinoma	6	1 renal carcinoma + leiomyosarcoma + liposarcoma	6	22
					1 lung spinocellular carcinoma	6
					1 pancreatic carcinoma	14
					1 non-Hodgkin's lymphoma	6
Abe et al, 2006 (7)	RA	147	IFX + MTX/100	Placebo + MTX/47	None	0	None	0	36
Antoni et al, 2005 (IMPACT) (8)	PsA	104	IFX/52	Placebo/52	None	0	None	0	50
Antoni et al, 2005 (IMPACT 2) (9)	PsA	200	IFX/100	Placebo/100	None	0	1 basal cell carcinoma	NA	24
Van den Bosch et al, 2002 (10)	AS	40	IFX/20	Placebo/20	None	0	None	0	12
Braun et al, 2002 (11)	AS	70	IFX/35	Placebo/35	None	0	None	0	12
Van der Heijde et al, 2005 (12)	AS	279	IFX/201	Placebo/78	None	0	None	0	24
Marzo-Ortega et al, 2005 (13)	AS	42	IFX + MTX/28	Placebo + MTX/14	None	0	None	0	30
Weinblatt et al, 1999 (14)	RA	89	ETN + MTX/59	Placebo + MTX/30	None	0	None	0	24
Moreland et al, 1999 (15)	RA	234	ETN/154	Placebo/80	None	0	None	0	24
Bathon et al, 2000 (16)	RA	632	ETN + placebo/415	MTX + placebo/217	1 breast cancer	NA	1 colon cancer	NA	52
					1 lung cancer	NA	1 bladder cancer	NA
					1 carcinoid lung cancer	NA
					1 Hodgkin's disease	NA
					1 prostate cancer	NA
Klareskog et al, 2004 (17)	RA	682	ETN + placebo/223	Placebo + MTX/228	2 basal cell carcinomas	NA	1 basal cell carcinoma	NA	54
			ETN + MTX/231		1 breast cancer	NA
					1 rectal cancer	NA
					1 melanoma	NA
Keystone et al, 2004 (18)	RA	420	ETN + MTX + placebo/367	Placebo + MTX/53	1 prostate cancer	NA	None	0	16
Van der Heijde et al, 2006 (19)	RA	503	ETN/163	MTX/152	2 gastric cancers	NA	1 breast cancer	NA	54
			ETN + MTX/188		1 lung cancer	NA
					1 basal cell carcinoma	NA
Combe et al, 2006 (20)	RA	254	ETN/103	SSZ/50	1 squamous skin cancer†	NA	None	0	24
			ETN + SSZ/101		1 acute myelogenous leukemia‡	12
Weisman et al, 2007 (21)	RA	535	ETN/266	Placebo/269	2 basal cell carcinomas	NA	1 carcinoma (NOS)	NA	16
							1 lung cancer	NA
							1 basal cell carcinoma	NA
Mease et al, 2000 (22)	PsA	60	ETN/30	Placebo/30	None	0	None	0	12
Mease et al, 2004 (23)	PsA	205	ETN/101	Placebo/104	None	0	None	0	24
Gorman et al, 2002 (24)	AS	40	ETN/20	Placebo/20	None	0	None	0	16
Brandt et al, 2003 (25)	AS	30	ETN/14	Placebo/16	None	0	None	0	24
Davis et al, 2003 (26)	AS	277	ETN/138	Placebo/139	None	0	None	0	24
Calin et al, 2004 (27)	AS	84	ETN/45	Placebo/39	None	0	None	0	12
Van der Heijde et al, 2006 (28)	AS	356	ETN/305	Placebo/51	None	0	None	0	12
Furst et al, 2003 (29)	RA	636	ADA + previous DMARDs/318	Placebo + previous DMARDs/318	1 T cell lymphoma§	8	None	0	24
					3 basal cell carcinomas¶	3, 10, 19
Weinblatt et al, 2003 (30)	RA	271	ADA + MTX/209	Placebo + MTX/62	1 colon cancer	18¶	None	0	24
Keystone et al, 2004 (31)	RA	619	ADA + MTX/419	Placebo + MTX/200	1 T cell lymphoma	9¶	1 basal cell carcinoma	24	52
					1 breast cancer	43¶
					1 gastrointestinal cancer	14¶
					1 seminoma	8¶
					3 basal cell carcinomas	8, 22, 27¶
					1 squamous cell carcinoma	28¶
Van de Putte et al, 2004 (32)	RA	544	ADA/434	Placebo/110	1 cholangiocarcinoma	2¶	1 basal cell carcinoma	6	26
					1 gastric cancer	9¶
					1 squamous cell carcinoma	7¶
					1 basal cell carcinoma	20¶
					1 lymphoma (MALT)	102¶
Breedveld et al, 2006 (33)	RA	799	ADA/274	Placebo + MTX/257	1 ovarian cancer	NA	1 lymphoma	NA	104
			ADA + MTX/268		1 breast cancer	NA	1 melanoma	NA
					1 colon cancer	NA	1 prostate cancer	NA
					1 metastatic cancer	NA	1 breast cancer	NA
					1 prostate cancer	NA
					1 multiple myeloma	NA
Mease et al, 2005 (34)	PsA	315	ADA/153	Placebo/162	None	0	None	0	24
Van der Heijde et al, 2006 (35)	AS	315	ADA/208	Placebo/107	None	0	None	0	24
Emery et al, 2008 (COMET) (36)	RA	542	ETN + MTX/274	MTX/268	1 chronic lymphatic leukemia	NA	3 breast cancers	NA	24
					1 epidermoid cancer of the tongue	NA	1 prostate cancer	NA
					1 basal cell carcinoma	NA
					1 Bowen disease	NA

Retrospective personal case series analysis results.

Study patients.

A total of 369 patients with RA, PsA, and AS were considered suitable candidates to receive anti-TNF drugs. Of these, 6 patients (1.6%) were excluded on the basis of exclusion criteria other than those for cancer screening. Therefore, 363 patients were eligible for this study.

Controls.

Of 75 controls, 73 (49 with RA, 9 with PsA, and 15 with AS) requiring anti-TNF therapy were treated. Two patients (2.7%) with RA were excluded due to history of tuberculosis in one case and recurrent respiratory infection in the other.

The demographic and clinical characteristics of the 2 groups are shown in Table 2. The 2 groups were unbalanced because none of the controls had received adalimumab due to the unavailability of the drug during the years 1999–2001.

Table 2. Clinical characteristics and frequency of occult cancer and after-therapy cancer occurrence in 363 study patients and 73 controls*

	Study patients (2002–2006)	Controls (1999–2001)	P
* Values are the mean ± SD unless otherwise indicated. RA = rheumatoid arthritis; NS = not significant; PsA = psoriatic arthritis; AS = ankylosing spondylitis; anti-TNF = anti–tumor necrosis factor. † More than 20 cigarettes/day for more than 20 years. ‡ Adalimumab was not available from the years 1999–2001.
Overall screened patients, no.	363	75
RA, no. (%)	187 (51.5)	44 (58.7)	NS
PsA, no. (%)	98 (27)	16 (21.3)	NS
AS, no. (%)	78 (21.5)	15 (20.0)	NS
Age, years
Overall	52.5 ± 12.2	54.1 ± 7.2	NS
RA	58.3 ± 10.1	57 ± 7.8	NS
PsA	51.8 ± 8.3	52 ± 6.3	NS
AS	39.7 ± 10.6	35 ± 11.3	NS
Disease duration from diagnosis, years
RA	7.8 ± 5.72	8.4 ± 4.1	NS
PsA	6.7 ± 4.9	7.2 ± 6.7	NS
AS	9.5 ± 6.7	8.9 ± 9.4	NS
Patients age >50 years, no. (%)
Overall	226 (62.2)	48 (64.0)	NS
RA	151 (66.8)	34 (70.8)	NS
PsA	63 (27.9)	11 (22.9)	NS
AS	12 (5.3)	3 (6.2)	NS
Heavy smokers, no. (%)†	46 (12.7)	12 (16.0)	NS
Patients with abnormal cancer screening procedure, no. (%)	36 (9.94)	0	< 0.02
Occult cancer, no. (%)	4 (1.1)	0	NS
Patients treated with anti-TNF therapy
Overall, no.	359	73
RA, no. (%)	184 (51.3)	42 (57.5)	NS
PsA, no. (%)	97 (27)	16 (21.9)	NS
AS, no. (%)	78 (21.7)	15 (20.6)	NS
Women/men
Overall	209/150	45/28	NS
RA	143/41	32/10	NS
PsA	49/48	9/7	NS
AS	17/61	4/11	NS
Duration of anti-TNF therapy, months	37.72 ± 23.39	10.81 ± 14.12	< 0.05
Infliximab, no.	174	58	< 0.05
Duration of therapy, months	34.52 ± 20.65	14.34 ± 11.54
Etanercept, no.	161	15	< 0.02
Duration of therapy, months	29.95 ± 16.22	8.23 ± 4.21	< 0.05
Adalimumab, no.	112	0‡	< 0.001
Duration of therapy, months	25.15 ± 13.96	0‡
Switching anti-TNF therapy, no. (%)
Overall	72 (20.05)	15 (20.54)	NS
RA	48 (66.7)	7 (46.7)	NS
PsA	11 (15.3)	5 (33.3)	NS
AS	13 (18.0)	3 (20.0)	NS
Cancer occurrence, no. (%)	1 (0.27)	3 (4.10)	< 0.02
Type of cancer/interval from starting anti-TNF therapy, weeks	1 larynx carcinoma/292	1 colon carcinoma/6	0.017
		1 colon carcinoma/8
		1 lung cancer/16
Followup duration, months	40.9 ± 16.7	50.6 ± 18.1	NS

Of the 363 study patients, 36 (9.94%) had at least 1 abnormal cancer screening procedure finding; none of the controls had an abnormal cancer screening procedure finding (P < 0.02). A total of 226 (62.3%) of 363 patients were age >50 years, including 151 (66.8%) of 226 with RA, 63 (27.9%) of 226 with PsA, and 12 (5.3%) of 226 with AS. Abnormal findings on US were detected in 8 (2.2%) of 363 patients, elevation of ≥1 tumor markers in 9 (2.5%) of 363, chest radiographs in 2 (0.55%) of 363, and hemoccult test in 16 (4.4%) of 363. Lung CT was carried out in 46 (12.7%) of 363 patients. After proper diagnostic procedures, 32 (88.9%) of 36 patients could be treated with anti-TNF therapy.

Occult cancer was diagnosed in 4 (1.1%) of 363 patients before starting therapy. Of these, 1 patient had lung cancer detected by chest CT, 1 patient with a positive hemoccult test result had colon cancer, 1 patient had prostate carcinoma heralded by elevated prostate-specific antigen levels, and 1 patient had renal cancer disclosed by US.

One study patient (0.27%) and 3 controls (4.1%) developed cancer over the followup period, with a statistically significant difference (P = 0.017). The comparison between the total occurrences of cancer in the study group (4 occult cancers and 1 established cancer) and the 3 occurrences of cancer that were assessed in the control group did not demonstrate a significant statistical difference (P = 0.96). In the study group, a 64-year-old woman with RA developed a larynx carcinoma with a 53-month interval from the beginning of therapy. She was a heavy smoker since age 23 years and she had received infliximab for 23 months before switching to etanercept. Three controls developed cancer after 6, 8, and 16 weeks from the beginning of therapy. The mean ± SD followup duration was 40.9 ± 16.7 months for the study group and 50.6 ± 18.1 months for controls (P = 0.071).

DISCUSSION

A careful analysis of the data from the RCTs provides the opportunity for some considerations concerning the employed screening procedures for cancer. First, both in the active treatment and placebo arms, all but one malignancy occurred in patients with RA, suggesting an underlying disease-related risk. Second, RCTs do not show if the demographic and clinical characteristics of patients randomized to active treatment or placebo were balanced with respect to preexisting risk factors for cancer, including cigarette smoking, previous or concomitant therapies such as cytotoxic drugs or radiation therapy, site locations in multicenter studies due to the different incidences of cancer in different countries, and comorbidities. Third, a careful analysis of RCTs shows an unexpected short interval from study entry and onset of cancer in a considerable percentage of patients in both active treatment and placebo arms. This may raise the suspicion of possible methodologic defects, especially regarding the screening procedures for cancer. The adopted procedures seem rather loose, hence patients were excluded only on the basis of “current diagnosis of cancer or history of cancer over the previous 5–10 years” (1–36), regardless of other risk factors or predisposing conditions for malignancy development.

In the study by Combe et al (20), a patient known to have myelodysplastic syndrome developed acute myeloid leukemia within 12 weeks of study entry. In addition, in the trial by Furst et al (29), one patient developed T cell lymphoma 8 weeks after enrollment. However, he had highly suspicious symptoms for this malignancy, including loss of appetite, weight loss, and night sweats prior to the study entry. Fourth, the interval from study entry to the onset of cancer was not obtained in 22 (39.3%) of 56 cases of active treatment and in 11 (64.7%) of 17 patients in the placebo arms. However, on the basis of available data mainly related to RCTs of infliximab and adalimumab, 19 (26.0%) of 73 malignancies occurred within 12 weeks of the enrollment. Of these, 15 (26.8%) of 56 were recorded in the active treatment arms and 4 (23.5%) of 17 were recorded in the placebo arms. Although the pathobiology of cancer may be variable, this data seems to indicate that at least a part of these malignancies was present before starting therapy. The similar percentage of cancer detected within the first 12 weeks of treatment in both patients receiving active medications and in those taking placebo seems to reinforce this statement. Fifth, 13 (23.2%) of 56 cancers in active treatment arms and 5 (29.4%) of 17 in the placebo arms were basal cell carcinomas of the skin. However, in all RCTs, the current diagnosis of basal cell carcinoma did not constitute an exclusion criterion. Therefore, patients were not checked for this malignancy and some of the reported skin cancers probably preceded the treatment. Given the high percentage of patients with basal cell carcinomas in both the active treatment and placebo arms, it is worth emphasizing that a thorough generalized examination of the skin should be performed before initiating anti-TNF therapy.

The relationship between the onset of cancer detection and the initiation of anti-TNF therapy has been reviewed previously. In a meta-analysis from the Mayo Clinic (43), it was noted that 13 (37%) of the 35 cancers were detected within 10 weeks of the initiation of anti-TNF therapy. However, in 2 large database studies addressing the issue of malignancy and anti-TNF therapy (47, 64), there was no comment made about the timing of cancer detection in either study, neither of which made a compelling case for an increased risk of cancer (except skin cancer).

These observations suggest that the cancer screening procedures used in all of the RCTs of anti-TNFα medications seem quite loose and need to be revised, especially if we take into account the length of the preclinic, asymptomatic phase of some solid cancers, including lung, breast, gastrointestinal, and genitourinary malignancies. Therefore, beyond the current diagnosis or a history of cancer, additional procedures to detect occult cancers in patients requiring anti-TNFα therapy seem to be appropriate.

Our retrospective case series analysis seems to confirm this statement. We selected more comprehensive screening procedures, considering the well-known risk factors for malignancy and in keeping with those recommended by the American Cancer Society (59) for the general population to detect cancer in the preclinical phase, including fecal occult blood research and CA-19-9 antigen determination for gastrointestinal and hepatobiliary cancers (65); prostate-specific antigen determination for prostate carcinoma (58); carcinoembryonic antigen, CA-15-3 antigen, and cancer antigen 125 test for breast and ovarian cancers (66, 67); and β₂-microglobulin for hematologic malignancies (68). These tumor markers are known to have a high rate of false positive cases (69), but in the current medical practice they are analyzed in the clinical context and therefore they are useful to tailor the treatment. In addition, since the sensitivity of chest radiographs for early lung cancer detection is low, in patients age >50 years with a long history of cigarette smoking, we also performed a chest CT (70). Prior to the adoption of this set of procedures, in 73 controls we observed 3 solid cancers occurring a few weeks from the beginning of the treatment, suggesting their preexisting nature. By contrast, we detected 4 occult cancers in 363 study patients before starting therapy.

Chest CT was performed in 46 high-risk patients, and in one male patient this examination disclosed a 1-cm non–small cell lung carcinoma in an otherwise asymptomatic heavy smoker with normal chest radiograph results. Due to the frequent assumption of corticosteroids and NSAIDs, the screening performance of the fecal occult blood search in patients with inflammatory rheumatic disorders may produce lower results than in the general population (71, 72). We found positive hemoccult tests in 16 (4.4%) of 363 patients, and in one case, occult colon cancer was detected. Elevated tumor markers and abnormal US findings were recorded in 9 (2.48%) and 8 (2.2%) patients, respectively. Of these, one patient with elevated prostate-specific antigen levels had prostate cancer and another patient had renal cancer detected by US. Both patients were otherwise symptom-free.

Since HRTs may represent a risk factor for the development of estrogen-dependent malignancies (73, 74), in our set of cancer screening procedures we also included the assumption of these therapies and contraceptive pills, recommending only mechanical contraception. HRT in postmenopausal women did not represent an exclusion criterion in RCTs and no information is available about the comparison of treatment and placebo arms with respect to this risk factor. In addition, hormonal contraception is recommended in all anti-TNFα trials and in clinical practice. Of note, 7 (12.5%) of 56 cancers in the active treatment arms (6 breast cancers and 1 endometrial cancer) and 2 (11.8%) of 17 in the placebo arms (2 breast cancers) were estrogen-dependent cancers. In our cohort of patients treated with anti-TNFα agents, no estrogen-dependent cancer occurred over prolonged exposure to therapy.

In conclusion, after we applied more comprehensive cancer screening procedures for patients requiring anti-TNFα medications over a 5-year period, we observed only 1 malignancy (0.27%) in 359 patients, with a significant difference with respect to 3 cancers (4.1%) diagnosed in 73 controls. Notably, these procedures allowed detection of occult cancer in 4 asymptomatic patients who would not have been excluded if they were screened only on the basis of cancer or history of cancer.

This study has several limitations. Because of the retrospective design, limited information is available regarding the direct assessment of the occurrence of malignancies and the effect of possible concomitant treatment. The use of an historical control group instead of a concurrent age- and sex-matched control group makes it difficult to compare the 2 groups due to the difference in the quality of the laboratory test used in the 2 distinct periods of time.

Even if it is not possible to draw firm conclusions, our results and those of RCTs showing 26.0% of malignancies diagnosed within 12 weeks from the beginning of therapy seem to indicate the need for a revision of the current cancer screening procedures in RCTs and in clinical practice. It would be desirable to confirm these results in a larger prospective multicenter randomized trial.

AUTHOR CONTRIBUTIONS

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Nannini had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Nannini, Cantini, Niccoli, Cassarà, Salvarani, Olivieri, Lally.

Acquisition of data. Nannini, Cantini, Niccoli, Cassarà, Salvarani, Olivieri.

Analysis and interpretation of data. Nannini, Cantini, Salvarani, Olivieri, Lally.