To the Editor:

Type I collagen (CI), the most abundant protein in humans, may play a pivotal role in the pathogenesis of systemic sclerosis (SSc) (1–6). Oral tolerance studies in rodents have shown that the dose-response curve for orally administered antigen is bimodal, with tolerance being induced optimally using low and high doses of antigen (7). It has been postulated that high-dose oral antigen induces predominantly clonal deletion of antigen-specific T cells, while low-dose oral antigen induces regulatory T cells (8, 9). Work from our institution has demonstrated that oral administration of CI at 500 μg/day in patients with SSc induces tolerance, as characterized by significant reductions in interferon-γ (IFNγ) and interleukin-10 (IL-10) production by peripheral blood mononuclear cells (PBMC) cultured with α1(I) and α2(I). This dosage of CI also effected improvement in several disease parameters, including skin scores (modified Rodnan skin scores [MRSS]) (10), patient assessment of disease activity (modified Health Assessment Questionnaire [M-HAQ]) (11), and results of pulmonary function tests (PFTs), including diffusing capacity for carbon monoxide (DLCO) and forced vital capacity (FVC) (12).

Studies in humans with rheumatoid arthritis (9), as well as in animals (13, 14), suggest that the oral dosage of an autoantigen such as collagen is pivotal in determining whether tolerance occurs and disease activity is suppressed. In our previous study of oral CI administration to patients with SSc, we used high doses of collagen (500 μg/day) and did not explore the effects of lower-dose regimens (12). The purpose of the present study was to determine whether lower doses of oral CI (10 μg/day and 100 μg/day) would induce immune tolerance to CI in patients with SSc, and/or whether a change in clinical parameters could be effected after 5 months of administration of CI at either or both of these dosages.

The study population consisted of 4 men and 18 women with SSc (17 white, 5 African American). Patients were recruited from the University of Tennessee (UT) and community rheumatology practices in Memphis and surrounding areas. The mean ± SD duration of disease in the study population was 6.7 ± 7.5 years (range 1 month to 30 years). Eight patients had limited SSc and 14 had diffuse SSc. Written consent for participation in the study was obtained from patients in accordance with the Helsinki II declaration, and the protocol was approved by the UT Institutional Review Board. The study was conducted under US Food and Drug Administration Investigational New Drug application 6575. The inclusion criterion was a clinical diagnosis of limited or diffuse SSc made by the study physician, according to the criteria of the American College of Rheumatology (formerly, the American Rheumatism Association) (15). Exclusion criteria were similar to those in our previous study (12); however, treatment with nonsteroidal antiinflammatory drugs (NSAIDs) was not allowed. The rationale for the exclusion of NSAIDs was based on work from our institution suggesting that NSAIDs may impede the development of oral tolerance (16).

For the first 5 months, all patients received placebo (0.1M acetic acid). Patients were then randomized to receive CI at 10 μg/day or 100 μg/day for 5 months. Those who had received the 10 μg/day CI dosage for 5 months were then crossed over to receive the 100 μg/day dosage for 5 months, and vice versa. Collagen was solubilized in 0.1M acetic acid and aliquoted into individual-dose vials which the patients kept refrigerated. Each morning, the patient added 1 vial of the CI preparation to 4–6 ounces of orange juice, and this was consumed before breakfast. Compliance was monitored by counting the milliliters of study drug left in vials returned at each visit. Patients were considered to have complied with the protocol if they have consumed at least 70% of the study medication. The mean consumption was 88.8% (range 71.2–96.9%).

All assessments were performed at 0, 5, 10, and 15 months. The MRSS (assessed by the same trained physician throughout) and the M-HAQ score were recorded. PFTs, including FVC and DLCO studies, were performed by the same technician (under the direction of a pulmonologist), using the same equipment each time. IFNγ and IL-10 levels were measured, by commercial enzyme-linked immunosorbent assay (ELISA; R&D Systems, Minneapolis, MN), in supernatants that were harvested from in vitro cultures of SSc PBMCs and stimulated with phytohemagglutinin (PHA; 5 μg/ml), α1(I) (50 μg/ml), or α2(I) (50 μg/ml) for 6 days. IFNγ and IL-10 levels in unstimulated PBMC culture supernatants were subtracted at each time point from the levels in supernatants from PBMCs stimulated with PHA, α1(I) or and α2(I). Serum levels of soluble IL-2 receptor (sIL-2R), a marker of in vivo immune activation, were measured by ELISA (R&D Systems). All samples were tested in duplicate.

Data were analyzed in a covariance structural model to fit a crossover design (PROC MIXED, version 9.1; SAS Institute, Cary, NC). Analyses were performed to determine whether there were significant differences in the mean responses to collagen at 10 μg/day versus 100 μg/day, collagen at 10 μg/day versus placebo, and collagen at 100 μg/day versus placebo. Results were adjusted for baseline measurements (including MRSS, M-HAQ score, PFT results, and cytokine levels), age, race, sex, disease duration, and type of SSc (limited versus diffuse).

Eighteen patients received 5 months of placebo treatment, 14 received at least 1 dose of oral CI, and 11 completed the entire study. The 11 study withdrawals were due to issues regarding compliance (n = 6), traumatic Colles fracture (n = 1), intercurrent medical illnesses not related to SSc (n = 2), pregnancy (n = 1 [during placebo treatment]), and allergic reaction to collagen characterized by a skin rash (n = 1). The final sample size was similar to that in our previous study, in which only 17 patients completed the protocol, and of these, only 11 actually had PFTs performed (12). Despite these small numbers, statistically significant results had been demonstrated for immune and clinical parameters, including PFT results, after 1 year of treatment with oral bovine CI at 500 μg/day (12).

There were no significant differences in response to CI at either 10 μg/day or 100 μg/day compared with placebo, with respect to clinical parameters including MRSS, M-HAQ score, and PFT results (FVC and DLCO). Interestingly, there was a significant increase in IFNγ production in response to α2(I) by PBMCs from patients receiving CI at 100 μg/day compared with both the placebo group (P = 0.02) and the 10 μg/day CI group (P = 0.01). There were no significant differences in the IFNγ response to α1(I) or in the IL-10 response to α1(I) or α2(I) among the treatment groups. In addition, levels of sIL-2R did not differ among the groups (P > 0.05).

The major finding of our study was that there was no significant response in any of the clinical parameters evaluated, including MRSS, M-HAQ, or PFTs, to oral bovine CI administered at either 10 μg/day or 100 μg/day for 5 months. Furthermore, there was no suppression of the T cell response to CI (i.e., reduction in IFNγ and/or IL-10 production) and no change in systemic immune activity. However, with the 100 μg/day dosage of CI, we did note increased in vitro production of IFNγ; this response was seen with α2(I) stimulation, but not α1(I) stimulation. Increased IFNγ production by peripheral lymphoid cells cultured with tolerizing antigen has been associated with induction of tolerance to some orally administered antigens in humans and mice (17, 18). Whether this enhanced IFNγ production in response to α2(I) represents tolerance induction or whether increased production of endogenous IFNγ is beneficial in SSc requires further investigation. Induction of IFNγ may be important because of the antifibrotic properties of this cytokine (19).

It should be noted that our results are limited by the small number of patients who completed the study, the wide range of disease duration, and the probable need to administer oral CI for longer than 5 months to observe a change in skin scores or pulmonary function. However, our findings suggest that low-dose CI may have some effect on cytokine profiles in SSc, although no response in any clinical parameters was noted.

Laura D. Carbone MD*, Kevin McKown MD†, Muthiah Pugazhenthi MD‡, Karen D. Barrow MS‡, Kenneth Warrington MD‡, Grant Somes PhD‡, Arnold E. Postlethwaite MD‡, * University of Tennessee, Health Science Center, Memphis, TN, † University of Wisconsin, Madison, WI, ‡ University of Tennessee, Health Science Center, Memphis, TN.