Minocycline is not effective in systemic sclerosis: Results of an open-label multicenter trial

Abstract

Objective

To determine if minocycline therapy improved skin thickness in early, diffuse systemic sclerosis (SSc) by ≥30%, a level of improvement unlikely to occur in the natural history of the disease as determined by recent controlled trials.

Methods

Subjects with diffuse SSc of ≤5 years' duration were treated with oral minocycline for 1 year. The primary outcome measure was the modified Rodnan skin thickness score (MRSS).

Results

Of 36 subjects initially enrolled, 31 returned for at least 1 followup visit and were included in the analysis (modified intent-to-treat analysis). The group consisted of 23 women and 8 men, with a mean age of 51.7 years (range 26–82 years) and a mean disease duration of 23.5 months (range 6–60 months). The mean MRSS at entry was 22.7 (range 12–43), and at the final visit it was 18.6 (range 2–48). There was no statistically significant difference in the change in skin scores between the minocycline-treated subjects and subjects previously reported in the D-penicillamine (D-Pen) trial. In addition, when adjusted for disease duration, a comparison of MRSS in the minocycline trial subjects (including all subjects active at each time point) and the previously reported D-Pen trial subjects showed no difference and no treatment effect.

Fourteen subjects did not complete all 12 months of treatment; 10 of them withdrew due to disease progression. Disease duration was significantly shorter for the noncompleters than for the completers (P < 0.03).

Conclusion

The degree of change in the MRSS was similar to that expected in the natural course of this disease. Based on these data, minocycline is not an effective therapy for SSc.

Minocycline was reported to be effective in the treatment of systemic sclerosis (SSc) in an open-label study involving 9 patients (1). The current study was designed to test this agent in a larger number of patients and, if a positive effect was observed, to serve as the rationale for a controlled trial to determine effectiveness. A positive effect was defined a priori as a 30% or greater improvement in the modified Rodnan skin thickness score (MRSS) (2). This effect magnitude was chosen on the basis of 2 earlier controlled trials in diffuse SSc (3, 4) that demonstrated an ∼20% improvement in skin scores in the control group. In the D-penicillamine (D-Pen) trial (3), there was no significant difference between the control arm (low dose) and the treatment arm (high dose), and thus, the reported improvement was considered to be due to the natural history of skin thickness regression as measured by trained observers in the context of a clinical trial.

In the initial placebo-controlled relaxin trial (4), MRSS in the placebo group improved by 18.4% over the 6-month study period. Although this study showed an apparent treatment effect of low-dose relaxin compared with placebo, a second placebo-controlled trial did not confirm this observation (5). For the current study of minocycline, a treatment effect of 30% improvement in MRSS was chosen as likely to represent a clinically significant change that could not be attributed to the natural history of the disease. Data from the D-Pen trial were used as an estimate of the expected skin score change in a comparator group.

PATIENTS AND METHODS

Study design.

This trial was a 3-center, open-label, 12-month-per-patient study of minocycline (50 mg twice a day for 1 month, then 100 mg twice a day for 11 months) for the treatment of patients with diffuse cutaneous scleroderma. The primary outcome measure was a change in the mean MRSS from first visit to end of study as compared with the D-Pen comparator group. Secondary outcomes included change in means of active hand extension and oral aperture values from study entry to end of study.

Patients had to meet American College of Rheumatology (formerly, the American Rheumatism Association) preliminary classification criteria for SSc (6), had to have diffuse cutaneous involvement (skin thickening proximal to the elbow and/or knee) with or without face and neck involvement (7), and an SSc duration of ≤60 months from the first SSc manifestation other than Raynaud's phenomenon.

Patients were excluded for the following reasons: age <18 years, pregnancy or lactation, allergy to tetracyclines, previous treatment with a tetracycline for >2 months since the onset of SSc, requirement for prolonged use of oral anticoagulants, presence of another well-defined rheumatic disease or only localized scleroderma (morphea or linear scleroderma) or scleroderma-like illnesses, serum creatinine ≥2 mg/dl, renal crisis in the 2 months prior to enrollment, other serious organ involvement (e.g., symptomatic pericarditis, uncontrolled malabsorption), or another chronic debilitating illness (e.g., cancer) which, in the opinion of the investigator, would require disallowed medication or make completion of the study unlikely.

Scleroderma hypertensive renal crisis was defined as the new onset of hypertension, accompanied by a rise in serum creatinine ≥50% over the patient's baseline or 120% over the upper limit of normal, with or without new proteinuria or microangiopathic changes on a peripheral blood smear (8). Scleroderma normotensive renal crisis was defined as an increase in serum creatinine, as noted above, plus the new onset of at least 1 of the following 5 features: proteinuria ≥2+ by dipstick, hematuria ≥2+ by dipstick or ≥10 red blood cells per high-power field, thrombocytopenia of <100,000 platelets/mm³, hemolysis with typical blood smear changes or increased reticulocyte count, or renal biopsy findings consistent with scleroderma renal crisis (8). Synovitis was defined as objective joint swelling (distinctive from diffuse hand or foot swelling) accompanied by tenderness on joint compression.

Corticosteroids, if taken, must have been at a stable dosage of ≤10 mg of prednisone (or equivalent) per day for at least 1 month prior to enrollment. All medications thought to be potentially disease modifying were excluded, including D-Pen, azathioprine, cyclophosphamide, methotrexate, chlorambucil, captopril, para-aminobenzoate potassium, mycophenolate mofetil, tamoxifen, etanercept, infliximab, leflunomide, and cyclosporine.

Brief courses of corticosteroids (>10 mg of prednisone or equivalent per day) were allowed as long as the duration of the increased dosage did not exceed 3 weeks at a time, and no more than 3 courses occurred in the 12-month study period.

Outcome measurements.

At study entry and every 3–4 months, patients were evaluated for the degree and extent of SSc skin thickening using the MRSS technique (9): skin thickness was assessed clinically in each of 17 body areas and scored using a 0–3 scale, where 0 = normal, 1 = mild thickness, 2 = moderate thickness, and 3 = severe thickness (maximum score 51). In addition, active handspread from the thumb to the fifth digit and maximum oral aperture (lip to lip) were measured in millimeters.

Study approval.

This study was approved by the Institutional Review Boards of Wayne State University, Medical College of Wisconsin, and the University of Connecticut.

Statistical analysis.

Statistical analyses were performed using SAS software (SAS Institute, Cary, NC). Continuous data were expressed as the mean ± SD; 95% confidence intervals (95% CIs) were calculated using standard methods. Group means comparing changes in the MRSS in this study with those in the D-Pen study were analyzed using Student's t-test (2-tailed, unpaired for comparisons across treatment groups). Secondary outcome measures of hand extension, as well as oral aperture measures between the baseline and the end of study (12 months or last observation) were compared using Student's t-test (2-tailed, paired for within-group analysis). In some cases, a modified intent-to-treat (ITT) analysis was used in which the last observation on a patient was used as the end-of-study value. For other analyses, outcome values of the minocycline completers and noncompleters were separately compared with the D-Pen study values. A univariate linear regression analysis of MRSS adjusted for disease duration was used on data from both the minocycline trial and the D-Pen trial.

The data set of the 68 completers in the D-Pen trial was obtained from the principal investigator (Dr. Philip Clements) at the data-coordinating center at the University of California, Los Angeles. Since the D-Pen trial enrolled individuals with 18 months or less of disease and had a 2-year-per-subject treatment period, outcome data regarding MRSS change scores for the second year of participation in the D-Pen trial were used for comparison with the minocycline trial.

RESULTS

Patient characteristics.

Thirty-six patients were enrolled in the study and 31 returned for at least 1 visit 3 months or more after study entry. There was no significant difference in demographic features or disease characteristics between the total group recruited (n = 36) and the 31 who were considered suitable for evaluation (data not shown). Of the 5 who were excluded from the analysis, 3 patients did not return for followup and 2 were withdrawn due to the institution of disallowed medications prior to the first return visit (methotrexate for psoriasis in 1 patient and 30 mg prednisone per day for synovitis in another).

The baseline characteristics of the minocycline study patients are shown in Table 1. Of the 31 patients suitable for evaluation (ITT patients), 23 were women, the mean age was 51.7 years (range 26–82 years), and the mean disease duration was 23.5 months (range 6–60 months). Antinuclear antibodies were positive in 93% of the patients and 32% were anti–Scl-70 positive. As expected for this group of patients with diffuse disease, no one was anticentromere antibody positive. Table 1 also shows disease characteristics of the 68 patients who completed months 12–24 in the D-Pen trial. Average disease duration was similar between the 2 groups, although the range was much greater in the minocycline trial.

Table 1. Baseline characteristics of all minocycline study patients, minocycline noncompleters, minocycline completers, and D-penicillamine (D-Pen) completers beginning the second year of the D-Pen study*

Feature	All minocycline (ITT) patients (n = 31)	Minocycline noncompleters (n = 14)	Minocycline completers (n = 17)	D-Pen completers (n = 68)†
* ITT = intent to treat; MRSS = modified Rodnan skin thickness score. † The data refer to the D-Pen study patients at month 12 in this 24-month study. ‡ P < 0.02 versus minocycline completers. § Three patients were not tested, and all 3 were in the minocycline completer group.
Sex, no. (%)
Female	23 (74)	8 (57)	15 (88)	51 (75)
Male	8 (26)	6 (43)	2 (12)	17 (25)
Race, no. (%)
White	23 (74)	10 (71.4)	13 (76.5)	44 (64.7)
Black	7 (23)	4 (28.6)	3 (17.6)	14 (20.6)
Other	1 (3)	0	1 (5.9)	10 (14.7)
Age, years
Mean ± SD	51.7 ± 13.9	50.1 ± 13.5	53.0 ± 14.5	45.2 ± 13.1
Range	26–82	27–76	26–82	20–70
Disease duration, months
Mean ± SD	23.5 ± 16.7	16.0 ± 11.6‡	29.8 ± 7.5	22.8 ± 4.4
Range	6–60	6–48	6–60	14–31
MRSS at entry
Mean ± SD	22.7 ± 7.7	23.7 ± 8.1	21.9 ± 7.5	17.0 ± 9.8
Range	12–43	12–43	12–37	2–44
Antinuclear antibody, % positive	93	–	–	–
Anticentromere antibody, % positive	0	0	0	–
Anti–Scl-70 antibody, no. (%) positive§	10 (35.7)	3 (21.4)	7 (50.0)	–

Table 2 shows the change in MRSS, the primary outcome measure, for the minocycline study population as well as for the D-Pen patients. Mean ± SD MRSS for the minocycline ITT patient group changed by −4.1 ± 7.19 (95% CI 1.46, 6.73), which was not significantly different from the MRSS change of −2.8 ± 13.7 (95% CI −0.58, 6.08) for the 68 D-Pen completers (P = 0.623). Although the 17 minocycline completers had a greater improvement in MRSS (from a mean ± SD of 21.9 ± 7.5 at entry to 14.6 ± 6.7 at the final visit, a difference of −7.3 ± 3.96 [95% CI 5.26, 9.33]) than the minocycline group as a whole, the change in scores between the 17 minocycline completers and the 68 D-Pen completers was not statistically significant (P = 0.195). As expected, the minocycline noncompleter group had minimal change in mean skin score from study entry (23.7 ± 8.1) to last observation (23.5 ± 9.9).

Table 2. Change in MRSS for minocycline study patients including ITT patients, minocycline noncompleters, and minocycline completers compared with D-penicillamine (D-Pen) completers in the second year of the D-Pen study*

Feature	All minocycline (ITT) patients (n = 31)	Minocycline noncompleters (n = 14)	Minocycline completers (n = 17)	D-Pen completers (n = 68)
* ITT = intent to treat; MRSS = modified Rodnan skin thickness score; 95% CI = 95% confidence interval. † At month 12. ‡ At month 24. § None of the differences between groups were statistically significant.
MRSS at study entry, mean ± SD (range)	22.7 ± 7.7 (12–43)	23.7 ± 8.1 (12–43)	21.9 ± 7.5 (12–37)	17.0 ± 9.8 (2–44)†
MRSS at final visit, mean ± SD (range)	18.6 ± 9.3 (2–48)	23.5 ± 9.9 (8–48)	14.6 ± 6.7 (2–26)	14.3 ± 5.0 (0–44)‡
Unit change in MRSS, mean ± SD (95% CI)§	−4.1 ± 7.19 (1.46, 6.73)	−0.2 ± 8.39 (−4.63, 5.06)	−7.3 ± 3.96 (5.26, 9.33)	−2.8 ± 13.7 (−0.58, 6.08)

Due to the tendency for skin scores to spontaneously improve over time, a comparison of MRSS in the minocycline patients (including all patients active at each time point) with that in the 68 D-Pen trial patients was done. When adjusted for disease duration, there was no significant difference between skin scores for the minocycline patients and those for the D-Pen patients.

In addition, none of the secondary outcome measures of right hand extension, left hand extension, or oral aperture changed significantly from baseline for the ITT group or for subgroups of completers (n = 17) or noncompleters (n = 14) (within-group comparison, data not shown).

Fourteen patients did not complete all 12 months of treatment. For these 14 noncompleters, the mean time in the study was 4.6 months, with 8 individuals withdrawing at 3–4 months, 5 at 6 months, and 1 at 9 months. Reasons for early withdrawal include the following. Ten had progression of disease including 1 or more of the following (note that each patient could have more than 1 worsening feature): progressive skin thickening with an increase in MRSS ≥15, to 117% of baseline value (n = 5), worsening lung involvement as evidenced by increasing dyspnea with a decline in forced vital capacity on pulmonary function testing, and/or changes of alveolitis on high-resolution computed tomography scan of the lungs (n = 2), onset of scleroderma renal crisis (n = 2), and synovitis requiring disallowed medication (n = 3). In addition, 1 patient withdrew consent due to lack of improvement in skin thickening (although no worsening) and unacceptable hyperpigmentation, 1 patient had a cerebrovascular accident (CVA), 1 was diagnosed as having lung cancer, and 1 was lost to followup.

Noncompleters differed significantly from completers in the duration of disease, with noncompleters having a shorter disease duration (mean ± SD 16.0 ± 11.6 months, range 6–48 months) compared with completers (29.8 ± 7.5 months, range 6–60 months) (P < 0.03). Although noncompleters were more likely to be men (43% of noncompleters versus 12% of completers), this difference was not statistically significant. Additionally, there was no statistically significant difference between these 2 groups in terms of race, age, MRSS at entry, or proportion positive for antinuclear antibodies or anti–Scl-70 (data not shown).

Adverse events.

Overall, minocycline was well tolerated. Excluding events that the investigators considered to be due to SSc disease progression (renal crisis [n = 2], development of or worsening of pulmonary fibrosis [n = 2], worsening gastrointestinal involvement [n = 8], digital ulcers [n = 1], new or worsening synovitis [n = 3]) and events thought by the investigators to be due to coincidental illness (1 each with CVA, lung cancer, and psoriasis), the following adverse events were considered likely to be drug related: lightheadedness or dizziness in 2 patients, oral or vaginal candidiasis in 2 patients, skin hyperpigmentation in 3 patients, and nausea and diarrhea in 6 patients. Only 1 patient withdrew from the study due to an adverse event (skin hyperpigmentation) attributed to study medication.

DISCUSSION

In this open-label trial of 31 patients, minocycline therapy was not shown to be better than the predicted natural history of skin involvement as demonstrated in the D-Pen trial. Open-label trials typically exaggerate, rather than underestimate, a treatment effect. This is particularly problematic in scleroderma studies using changes in the skin score as the primary outcome measure due to the fact that skin thickness typically progresses over the first 2–5 years of diffuse disease (10), then gradually improves in many individuals. Taking this natural or expected course of skin involvement into consideration, a mean decrease of 30% in MRSS was set a priori as the minimum change that would be interpreted as a positive result. This was based on recent controlled trials that used similar methodology with which these authors are experienced.

The high dropout rate of 45% (14 of 31 patients) clearly biases the results of the completer analysis. The improvement in MRSS in the 17 completers was better than in the ITT group, since most of those who did not complete the study (10 of 14) did so due to disease progression. In addition, the fact that the completers had a significantly longer mean disease duration than the noncompleters suggests that more of the completers were in the phase of spontaneous improvement.

Data from the high-dose versus low-dose D-Pen trial are particularly valuable here. The D-Pen trial did not demonstrate a difference in skin scores between the 2 arms and is widely interpreted as a negative trial, therefore providing data on the expected change of MRSS in a group of early, diffuse SSc patients. There are pitfalls in using such historic controls. The D-Pen study had a 2-year-per-patient treatment phase and enrolled patients with a disease duration of 18 months or less. To minimize differences in disease duration, data from the second year of the D-Pen trial were used as the comparison. As noted in Table 1, disease duration was similar between the year 2 D-Pen group and the minocycline group. However, patients who are midway in a 2-year trial may be very different from those who are newly recruited into a 1-year trial in terms of disease improvement or at least stability and multiple perceived other benefits. To address this issue, at least in part, a univariate linear regression analysis of MRSS adjusted for disease duration was used on data from both the minocycline trial (using data from all patients active at all available time points) and the D-Pen trial (using data from all 68 completers at all available time points). When this adjustment for disease duration was made, there was still no difference between the populations. In other words, skin scores in the minocycline group changed over time at the rate predicted by the D-Pen trial. This is strong evidence against a minocycline treatment effect.

Overall, minocycline was well tolerated. Only 1 subject dropped out of the study due to an adverse event (skin hyperpigmentation) attributed to the medication. Although the hyperpigmentation could have been due to scleroderma rather than the treatment, the patient was unwilling to continue.

In the absence of a concurrent control group, it is difficult to categorically conclude that the minocycline-treated group fared no better than an untreated group would have. However, the mean change in MRSS observed in this study was similar to the predicted change (based on recent trials with comparable methodology), leading the authors to conclude that additional controlled trials of this agent in SSc are not warranted.

The value of reporting a negative trial is to counterbalance the bias in the literature that favors the publication of positive studies over negative ones. The results of this trial are important not only to clinicians making treatment decisions, but also to clinical researchers who must determine the best direction for future trials. Additionally, it is crucial to emphasize that open-label trials in SSc must take into consideration the expected improvement in skin scores so as not to systematically overestimate a treatment effect. In this regard, the recent controlled trials in SSc comprise a source of invaluable information regarding the natural history of this disease for patients within a controlled trial setting. The authors remain optimistic that agents with a true treatment effect can be developed, tested, and shown to be effective in this most refractory of rheumatic diseases.

Acknowledgements

The authors wish to thank Dr. Philip Clements and the D-Penicillamine Trial Group for access to the data of the 68 trial completers. In addition, the authors thank Robert Lasky, PhD, Associate Professor, Center for Clinical Research and Evidence Based Medicine, University of Texas–Houston Health Science Center, for valuable statistical assistance.