Robert F. Spiera, M.D., Sebastian Unizony, M.D., Kenneth J. Warrington, M.D., Jennifer Sloane, M.D., Angeliki Giannelou, M.D., Michael C. Nivens, Ph.D., Bolanle Akinlade, M.D., Wanling Wong, Ph.D., Rafia Bhore, Ph.D., Yong Lin, M.D., Frank Buttgereit, M.D., Valerie Devauchelle‑Pensec, M.D., Ph.D., Andrea Rubbert‑Roth, M.D., George D. Yancopoulos, M.D., Ph.D., Frederic Marrache, M.D., Ph.D., Naimish Patel, M.D., and Bhaskar Dasgupta, M.D., for the SAPHYR Investigators*

BACKGROUND
More than half of patients with polymyalgia rheumatica have a relapse during tapering of glucocorticoid therapy. Previous studies have suggested that interleukin- 6 blockade may be clinically useful in the treatment of polymyalgia rheumatica. Sarilumab, a human monoclonal antibody, binds interleukin-6 receptor α and efficiently blocks the interleukin-6 pathway.

METHODS
In this phase 3 trial, we randomly assigned patients in a 1:1 ratio to receive 52 weeks of a twice-monthly subcutaneous injection of either sarilumab (at a dose of 200 mg) plus a 14-week prednisone taper or placebo plus a 52-week prednisone taper. The primary outcome at 52 weeks was sustained remission, which was defined as the resolution of signs and symptoms of polymyalgia rheumatica by week 12 and sustained normalization of the C-reactive protein level, absence of disease flare, and adherence to the prednisone taper from weeks 12 through 52

RESULTS
A total of 118 patients underwent randomization (60 to receive sarilumab and 58 to receive placebo). At week 52, sustained remission occurred in 28% (17 of 60 patients) in the sarilumab group and in 10% (6 of 58 patients) in the placebo group (difference, 18 percentage points; 95% confidence interval, 4 to 32; P = 0.02). The median cumulative glucocorticoid dose at 52 weeks was significantly lower in the sarilumab group than in the placebo group (777 mg vs. 2044 mg; P<0.001). The most common adverse events with sarilumab as compared with placebo were neutropenia (15% vs. 0%), arthralgia (15% vs. 5%), and diarrhea (12% vs. 2%). More treatment-related discontinuations were observed in the sarilumab group than in the placebo group (12% vs. 7%).

CONCLUSIONS
Sarilumab showed significant efficacy in achieving sustained remission and reducing the cumulative glucocorticoid dose in patients with a relapse of polymyalgia rheumatica during glucocorticoid tapering. (Funded by Sanofi and Regeneron Pharmaceuticals; SAPHYR ClinicalTrials.gov number, NCT03600818.)

From the Hospital for Special Surgery, Weill Cornell Medical College, New York (R.F.S.), and Regeneron Pharmaceuticals, Tarrytown (A.G., M.C.N., B.A., R.B., G.D.Y.) — both in New York; the Vasculitis and Glomerulonephritis Center, Massachusetts General Hospital, Harvard Medical School, Boston (S.U.), and Sanofi, Cambridge ( J.S., N.P.) — both in Massachusetts; the Division of Rheumatology, Mayo Clinic College of Medicine, Rochester, MN (K.J.W.); Sanofi, Bridgewater, NJ (W.W., Y.L.); the Department of Rheumatology and Clinical Immunology, Charité University Medicine, Berlin (F.B.); CHRU de Brest, Service de Rhumatologie, Brest (V.D.-P.), and Sanofi, Chilly- Mazarin (F.M.) — both in France; the Division of Rheumatology and Immunology, Cantonal Hospital St. Gallen, St. Gallen, Switzerland (A.R.-R.); and Anglia Ruskin University, Cambridge, United Kingdom (B.D.). Dr. Spiera can be contacted at spierar@ hss .edu or at Rheumatology, Hospital for Special Surgery, Belaire Bldg., 525 E. 71st St., 7th Fl., New York, NY 10021. *A list of the SAPHYR investigators is provided in the Supplementary Appendix, available at NEJM.org.

N Engl J Med 2023;389:1263-72. DOI: 10.1056/NEJMoa2303452 Copyright © 2023 Massachusetts Medical Society.

Polymyalgia rheumatica is an inflammatory disease of unknown cause that affects persons over the age of 50 years1 and is characterized by pain and morning stiffness of the shoulder and pelvic girdles, with significant effect on quality of life and function.^2-5 Glucocorticoids have been the mainstay of treatment in patients with this condition.^3,6-8 Although moderate doses of glucocorticoids (15 to 20 mg of prednisone) can control disease, more than half of patients cannot successfully taper glucocorticoid therapy. The resultant longterm treatment has been linked to substantial glucocorticoid-related side effects.^9-12 Adjunctive treatments, including methotrexate, have been investigated without clear demonstration of benefit.^8,13-16 Patients with polymyalgia rheumatica who have a relapse while tapering glucocorticoid therapy or have glucocorticoid-related adverse effects have limited treatment options.^6,17 Thus, treatments that hasten disease remission, decrease symptoms, reduce overall dependence on glucocorticoids and the associated toxic effects, and improve quality of life and function are needed.

Interleukin-6 has been implicated in the pathophysiology of polymyalgia rheumatica because circulating elevated levels and increased tissue expression of interleukin-6 have been identified in patients with this condition.^3,18 Studies have suggested that interleukin-6 blockade may be clinically useful in the treatment of polymyalgia rheumatica.^19-23 Biologic pathways independent of interleukin-6 probably also contribute to disease pathogenesis.^24,25

Sarilumab, a human monoclonal antibody, binds interleukin-6 receptor α and efficiently blocks the interleukin-6 pathway. We performed the Sarilumab in Patients with Polymyalgia Rheumatica (SAPHYR) trial to evaluate the efficacy and safety of sarilumab in patients with polymyalgia rheumatica who had a disease flare while they were tapering glucocorticoid therapy.

Trial Design and Oversight
SAPHYR was a phase 3, multicenter, randomized, double-blind, placebo-controlled trial involving patients who met the 2012 provisional classification criteria for polymyalgia rheumatica.²⁶ The trial was conducted at 60 sites in 17 countries in accordance with the principles of the Declaration of Helsinki and Good Clinical Practice guidelines and was approved by an independent ethics committee or institutional review board at each trial site. All the patients provided written informed consent.

The trial was sponsored by Sanofi and Regeneron Pharmaceuticals. The investigators and sponsors designed the trial and collected and analyzed the data. The sponsors provided sarilumab and placebo. All the authors participated in the interpretation of the data and in revisions to the manuscript and approved the submission of the manuscript for publication. All the authors vouch for the fidelity of the trial to the protocol and for the accuracy and completeness of the data and analyses reported.

Patients
All the patients had at least one episode of disease flare during a glucocorticoid taper (at a dose of ≥7.5 mg per day or prednisone dose equivalent) within 12 weeks before screening and had a history of at least 8 weeks of glucocorticoid treatment (≥10 mg per day or prednisone dose equivalent). All the patients had symptoms of polymyalgia rheumatica and an erythrocyte sedimentation rate (ESR) of at least 30 mm per hour or a C-reactive protein (CRP) level of at least 10 mg per liter within 12 weeks before screening. Excluded from the trial were patients who had received a diagnosis of giantcell arteritis, rheumatoid arthritis, or other inflammatory arthritis. Details regarding the inclusion and exclusion criteria are provided in the protocol, available with the full text of this article at NEJM.org.

Treatment
Patients were randomly assigned in a 1:1 ratio to receive 52 weeks of a twice-monthly subcutaneous injection of either sarilumab (at a dose of 200 mg) plus a 14-week prednisone taper or placebo plus a 52-week prednisone taper. This treatment was followed by a 6-week follow-up period. The dose of sarilumab could be reduced to 150 mg administered every 2 weeks in a blinded manner to manage neutropenia, thrombocytopenia, or elevated liver enzymes. Tapering doses of prednisone were administered in a blinded fashion at an initial dose of 15 mg per day for 2 weeks in the two groups to ensure adequate disease control at baseline (Table S1 in the Supplementary Appendix, available at NEJM .org). Patients were assessed at weeks 2, 4, 8, 12, 16, 20, 24, 32, 40, 52, and 58. Patients who did not have disease remission or who had disease flare within the initial 12 weeks were allowed to receive add-on prednisone (open-label at ≤5 mg per day) with the blinded taper. If add-on prednisone was discontinued before week 13, patients could continue to receive their assigned treatment. The administration of glucocorticoids was recommended for patients as rescue therapy for disease flare. The protocol specified the use of prednisone with a taper of 14 weeks in the sarilumab group and 52 weeks in the placebo group, whereas rescue therapy could be any glucocorticoid in either group, according to the investigator’s clinical judgment. Patients who were initiating glucocorticoid rescue therapy after week 12 had to discontinue the per-protocol blinded prednisone taper and were deemed to have had no response. Methotrexate (at a dose of ≤15 mg per week) was permitted if the dose had been stable for at least 3 months before baseline and thereafter throughout the trial.

Outcomes
The primary outcome at 52 weeks was sustained remission, which was defined as clinical remission (i.e., the resolution of signs and symptoms of polymyalgia rheumatica and normalization of the CRP level to <10 mg per liter) by week 12, along with the absence of disease flare, sustained CRP normalization, and adherence to the prednisone taper regimen from weeks 12 to 52. Disease flare was defined as the recurrence of symptoms or an elevation in the ESR from active polymyalgia rheumatica resulting in an increase in the glucocorticoid dose. Because sarilumab directly affects the CRP level, the measurement of CRP was performed in a blinded manner, and samples were centrally processed as a component of the sustained remission outcome. The measurement of the ESR was locally processed to aid investigators in their monitoring for potential disease flares. A dual-assessor approach was used to maintain blinding during the trial. The sponsor and the efficacy assessor were unaware of postbaseline CRP and ESR values to prevent the unmasking of the known effects of interleukin-6 receptor inhibition on these acute phase reactants. A safety assessor, the only investigator with access to laboratory data (except postbaseline CRP), was assigned at the trial site to perform a regular review of the patient’s laboratory data. An efficacy assessor determined the occurrence of a disease flare at each trial visit.

Key secondary outcomes were each of the components of the sustained remission outcome, the cumulative glucocorticoid dose during the 52-week period, the time until the first disease flare after clinical remission, the composite Glucocorticoid Toxicity Index score, and safety. (Scores on the Glucocorticoid Toxicity Index range from 0 to 538, with higher scores indicating more toxic effects.) Other outcomes were the disease activity score (PMR-AS), the score on the physician global assessment of disease activity– visual analogue scale (MD-VAS), and pharmacodynamics (i.e., change in the CRP level). Also evaluated were measures of quality of life and overall disability, including both the physical and mental component scores on the Medical Outcomes Study 36-Item Short-Form General Health Survey (SF-36), the score on the singleindex utility scale on the European Quality of Life–5 Dimensions (EQ-5D-3L) scale, the score on the fatigue scale on the Functional Assessment of Chronic Illness Therapy (FACIT-fatigue), and the score on the Health Assessment Questionnaire Disability Index (HAQ-DI). Details regarding these outcomes are provided in Tables S2 and S3.

Statistical Analysis
At the time of the trial initiation, the target enrollment was 280 patients. In July 2020, because of protracted recruitment exacerbated by the coronavirus disease 2019 (Covid-19) pandemic, the sponsor discontinued further enrollment. All the enrolled patients were permitted to complete the trial according to the protocol. Before the database lock, the statistical significance level was changed from 0.01 to 0.05 because of the reduced sample size. With an alpha level of 0.05, the enrollment of 59 patients in each group provided the trial with at least 85% power to detect a between-group difference in the primary outcome of 25 percentage points and at least 95 power to detect a difference of 30 percentage points. These power calculations assumed a response of 5 to 15% in the placebo group on the basis of a two-sided chi-square test with a significance level of 0.05.

The primary outcome of sustained remission at week 52 was summarized as counts and proportions in each group and was analyzed by Fisher’s exact test. We also calculated the difference between proportions and the 95% Wald asymptotic confidence limits. Patients were considered to have had no response if they did not have remission or if they had received rescue therapy with more than 5 mg of prednisone (or its equivalent) within 12 weeks after baseline, if they received any glucocorticoid rescue therapy during weeks 12 to 52 after baseline, or if they discontinued treatment before week 52. In the analyses in which median values were calculated, the data are reported with either simple range or interquartile range, as marked.

The majority of missing assessments were due to early treatment discontinuations, and patients with missing assessments were considered to have had no response. We imputed intermittently missing CRP values using the last-valuecarried- forward approach up to the last nonmissing measurement; we did not impute missing data for rescue therapy. We performed sensitivity analyses to handle missing data (using control-based multiple imputation and tipping-point analysis) for the primary outcome. Details regarding these analyses are provided in the Supplementary Appendix. Because of the known pharmacodynamic effect of interleukin-6 blockade on acute phase reactants, we performed a sensitivity analysis for the primary outcome that excluded measures of CRP and ESR.

For the cumulative glucocorticoid dose (including prednisone) over 52 weeks, the P value was calculated with the use of the nonparametric Wilcoxon rank-sum test. We used the Kaplan– Meier method to summarize the time until the first disease flare after clinical remission and used a descriptive summary for the proportion of patients with disease flare after clinical remission.

For the cumulative worsening score and aggregate improvement score for the Glucocorticoid Toxicity Index at week 52, we performed an analysis of covariance (using the treatment group and baseline glucocorticoid toxicity score as fixed effects) to test the least-squares mean between-group difference. Changes in scores on the PMR-AS and quality-of-life measures from baseline to week 52 were analyzed with mixedmodel repeated measures.

Multiplicity was controlled by a hierarchical testing procedure with an overall type I error rate of 0.05 (two-sided) for the primary outcome (tested first) and the secondary outcome of the total cumulative glucocorticoid dose (tested if the primary outcome reached 0.05 significance). There was no multiplicity adjustment for the other secondary outcomes. Results are reported with unadjusted 95% confidence intervals, which should not be used to infer definitive treatment effects.

Patients
From October 2018 through July 2020, a total of 118 patients underwent randomization (60 to the sarilumab group and 58 to the placebo group). Of these patients, 117 received treatment; 1 patient in the sarilumab group did not receive the assigned drug. A total of 78 patients completed treatment (42 in the sarilumab group and 36 in the placebo group). The most common reason for trial discontinuation was adverse events in the sarilumab group (7 patients) and lack of efficacy in the placebo group (9 patients) (Fig. S1).

The patients’ characteristics in the two groups were generally similar at baseline and were representative of the population with polymyalgia rheumatica. The median age of the patients was 69 years, 70% of the patients were women, and 83% were White (Table 1 and Tables S4 and S5). The overall median interval since the diagnosis of polymyalgia rheumatica was 300 days (simple range, 66 to 3992).

Sustained Remission
At week 52, sustained remission occurred in 28% of the patients (17 of 60) in the sarilumab group and in 10% (6 of 58) in the placebo group (difference, 18 percentage points; 95% confidence interval [CI], 4 to 32; P = 0.02). In a sensitivity analysis that excluded CRP and ESR values, the sustained remission was 32% (19 of 60 patients) in the sarilumab group and 14% (8 of 58 patients) in the placebo group at week 52

Table 1. Characteristics of the Patients at Baseline.*

* Plus–minus values are means ±SD. Median values are presented with either a simple range or an interquartile range (IQR), as marked.
† Race was reported by the patients. The collection of data regarding race or ethnic background was not permitted in France. No patients identified as Black or African American, Native Hawaiian or Other Pacific Islander, or American Indian or Alaska Native.
‡ The disease duration was measured from the date of diagnosis to the trial baseline. Data regarding disease duration were available for 54 patients in the sarilumab group and for 50 in the placebo group. The overall median disease duration in the two groups combined was 300 days (simple range, 66 to 3992).
§ Other immunosuppressive drugs were leflunomide (in 2 patients in the sarilumab group and 1 patient in the placebo group), hydroxychloroquine (in 1 patient in each group), azathioprine (in 1 patient in the placebo group), and adalimumab (in 1 patient in the sarilumab group).
¶ Included in this category are patients who reported having morning stiffness. ‖ On the physician’s global assessment of disease activity on the basis of a visual-analogue scale, scores range from 0 to 100, with higher scores indicating greater disease activity.
** Scores on the Glucocorticoid Toxicity Index range from 0 to 538, with higher scores indicating more toxic effects.

Figure 1. Sustained Remission and Its Components.

Shown is the proportion of patients with sustained remission at week 52 (primary outcome) in the sarilumab group and the placebo group. Remission is shown according to the individual components of the composite outcome (at left) and according to the overall sustained remission and an analysis of sustained remission with the exclusion of acute phase reactants (C-reactive protein [CRP] and erythrocyte sedimentation rate [ESR]) at week 52 (at right). Sustained remission at week 52 was defined as the resolution of signs and symptoms of polymyalgia rheumatica and CRP normalization (<10 mg per liter) by week 12, absence of disease flare from weeks 12 to 52, sustained CRP normalization from weeks 12 to 52, and adherence to the assigned prednisone taper from weeks 12 to 52.

(difference, 18 percentage points; 95% CI, 3 to 33). The results from the primary analyses and sensitivity analyses for missing data were consistent and supported the findings of the primary outcome (see the Supplementary Appendix). All the individual components of sustained remission favored the sarilumab group (Fig. 1). Among the patients who received concomitant methotrexate, sustained remission occurred in 3 of 12 (25%) in the sarilumab group and in 2 of 17 (12%) in the placebo group at week 52 (Table S6). The proportion of patients with no signs or symptoms of polymyalgia rheumatica was 57% (31 of 54) in the sarilumab group and 49% (26 of 53) in the placebo group at week 16; at week 52, the proportion of patients was 81% (39 of 48) and 57% (26 of 46), respectively (Fig. S2). With the exclusion of the patients who had received rescue therapy, this proportion at week 52 was 84% (27 of 32) and 47% (8 of 17) (Fig. S3). At 52 weeks, the proportion of patients who had no signs or symptoms and who had not received rescue therapy was 45% (27 of 60) in the sarilumab group and 14% (8 of 58) in the placebo group, despite a more rapid prednisone taper in the sarilumab group (14 vs. 52 weeks); among the evaluable patients at week 52 who had completed treatment, the proportion was 64% (27 of 42) in the sarilumab group and 22% (8 of 36) in the placebo group (Fig. S4). At the end of the 6-week post-treatment follow- up period, the proportion of patients who did not have active polymyalgia rheumatica among those who did not receive rescue therapy was higher in the sarilumab group than in the placebo group: 73% (22 of 30 patients) and 40% (6 of 15 patients), respectively. Additional details regarding the sustained remission data are provided in the Supplementary Appendix.

Cumulative Glucocorticoid Dose
By week 52, the median cumulative glucocorticoid dose was 777 mg (interquartile range, 777 to 1018) in the sarilumab group and 2044 mg (interquartile range, 1950 to 2840) in the placebo group (P<0.001). The median of the difference between the actual and expected glucocorticoid cumulative dose on the basis of the assigned prednisone taper was 0 mg in the sarilumab group and 199 mg in the placebo group. More than 50% of the patients in the sarilumab group had the same actual and expected cumulative glucocorticoid dose.

Disease Flare and Rescue Therapy
Clinical remission at 12 weeks occurred in 41 of 60 patients (68%) in the sarilumab group and in 30 of 58 patients (52%) in the placebo group. Disease flare after clinical remission occurred in 10 of 41 patients (24%) in the sarilumab group and in 17 of 30 patients (57%) in the placebo group (Fig. S5). During treatment, rescue therapy was administered in 19 of 60 patients (32%) in the sarilumab group and in 34 of 58 patients (59%) in the placebo group (Fig. S6). During the 6-week follow-up period, rescue therapy was administered in 15 of 45 patients (33%) in the sarilumab group and in 23 of 38 patients (61%) in the placebo group.

Glucocorticoid Toxicity Index and Disease Activity
On the Glucocorticoid Toxicity Index, the leastsquares mean change from baseline in the score for cumulative disease worsening during the 52- week treatment period was 52.32 in the sarilumab group and 57.22 in the placebo group (difference, −4.90; 95% CI, −23.48 to 13.67); the least-squares mean change in the aggregate improvement score was −4.02 and 2.57, respectively (difference, −6.59; 95% CI, −24.05 to 10.86). In the bone health domain, a loss in bone mineral density of more than 3% was observed in 10 of 46 patients (22%) in the sarilumab group and in 18 of 37 (49%) in the placebo group.

* Terms for adverse events are from the Medical Dictionary for Regulatory Activities, version 24.0. Adverse events were identified by the investigators and monitored by the sponsor team. If an adverse event did not appear to meet the criteria for an adverse event of special interest or a serious adverse event, the sponsor team queried the investigator to ensure adherence to protocol. A similar query process was followed for adverse evens of special interest or serious adverse events that clearly met the criteria according to the protocol but were not identified as such by the investigator.
† In the sarilumab group, adverse events that led to treatment discontinuation were reported in 1 patient each with coronavirus disease 2019 (Covid-19), intervertebral discitis, pneumonia, and nephrolithiasis and in 3 patients with neutropenia. In the placebo group, these events were reported in 1 patient each with Covid-19, giant-cell arteritis, pulmonary embolism, and increased alanine aminotransferase level. No deaths occurred in either trial group.

In the disease activity score, the least-squares mean change from baseline was −15.57 in the sarilumab group and −10.27 in the placebo group (difference, −5.30; 95% CI, −8.01 to −2.60) at week 52. More patients in the sarilumab group had a low disease activity score (<7) than in the placebo group (Table S7).

Change in CRP
At week 52, the mean CRP level decreased from baseline by 6.9 mg per deciliter in the sarilumab group and by 1.7 mg per deciliter in the placebo group. This decrease was greater in the sarilumab group at all time points that were measured (Fig. S8).

Safety
The most common adverse events during treatment (reported in ≥10% of the patients in the sarilumab group) were neutropenia (15% vs. 0 in the placebo group; unadjusted P<0.05), arthralgia (15% vs. 5%), diarrhea (12% vs. 2%), insomnia (10% vs. 16%), hypertension (10% vs. 3%), and osteoarthritis (10% vs. 9%). Adverse events that led to permanent treatment discontinuation — which occurred in 7 of 59 patients (12%) in the sarilumab group and in 4 of 58 patients (7%) in the placebo group — were driven primarily by a higher incidence of neutropenia in the sarilumab group. No infections were associated with neutropenia. In 1 patient, the dose of sarilumab was reduced to 150 mg every 2 weeks owing to neutropenia, which was mild in severity and resolved within 22 days. No deaths were reported in either group (Table 2). Serious adverse events, adverse events of special interest, and neutropenia according to grade are provided in Tables S8, S9, and S10. Posttreatment adverse events were reported in 4 of 59 patients (7%) in the sarilumab group and in 7 of 58 patients (12%) in the placebo group. Post-treatment serious adverse events were reported in 2 of 59 patients (3%) in the sarilumab group and in 1 of 58 (2%) in the placebo group. Additional details regarding post-treatment events are provided in the Supplementary Appendix.

In this phase 3 trial involving patients with polymyalgia rheumatica, significantly more patients had sustained remission at week 52 with sarilumab plus a 14-week prednisone taper than with placebo plus a 52-week prednisone taper. The results of a sensitivity analysis that excluded acute phase reactants (i.e., CRP and ESR) were consistent with the primary findings, which indicated that a reduction in CRP levels alone did not drive the observed effect with sarilumab.

The baseline clinical and demographic characteristics were generally similar in the two groups, but more patients were receiving methotrexate in the placebo group. The proportion of patients with sustained remission remained similar to the primary results in each group regardless of concomitant methotrexate use.

At week 52, the proportion of patients who had no signs or symptoms of polymyalgia rheumatica and who were not receiving either prednisone or rescue therapy was 45% in the sarilumab group and 14% in the placebo group.

In our trial, we recruited patients with polymyalgia rheumatica who had a relapse while tapering glucocorticoid therapy at a dose of 7.5 mg or more. All the patients had a median disease duration of 300 days and were receiving a median glucocorticoid dose of 10 mg per day at baseline. We considered that these patients had a response to therapy if they met a stringent composite primary outcome for clinical remission by week 12, which was defined as the resolution of signs and symptoms of polymyalgia rheumatica by week 12 and sustained normalization of the CRP level, absence of disease flare, and adherence to the prednisone taper from week 12 through 52. Additional evidence of the efficacy and safety of interleukin-6 blockade in patients with polymyalgia rheumatica has been provided in two recent trials of tocilizumab: PMR-SPARE,²² a 16-week trial involving 39 patients with new-onset polymyalgia rheumatica, and SEMAPHORE,²³ a 24-week trial involving 101 patients who had a relapse while tapering their glucocorticoid dose to less than 10 mg. These studies showed efficacy with respect to their primary end points. However, both trials had a shorter duration than our trial and did not show improvements in quality of life in patients receiving tocilizumab. In the SEMAPHORE trial, the primary end point did not include complete discontinuation of glucocorticoid treatment, and 51% of the patients receiving tocilizumab were still receiving glucocorticoids at the week 24 efficacy analysis, with a mean cumulative dose of 1380 mg.²³

During the 52 weeks of our trial, the cumulative glucocorticoid dose in the sarilumab group (777 mg) was lower than that in the placebo group (2044 mg), a disparity that was driven by protocol-directed between-group differences in the tapering regimen. Despite a faster prednisone taper in the sarilumab group than that recommended by guidelines,⁶ the need for rescue glucocorticoid therapy was lower than that in the placebo group and fewer patients had a disease flare after having clinical remission. These findings indicate a glucocorticoid-sparing and remission-maintenance effect of sarilumab.

No substantial between-group differences were observed in the scores on the Glucocorticoid Toxicity Index at week 52. However, it should be noted that the Glucocorticoid Toxicity Index has not been assessed or validated in patients who are being treated with the lower doses of glucocorticoids that are used in those with polymyalgia rheumatica.²⁷ Although our trial did not include formal hypothesis testing for quality-of-life outcomes, the results are consistent with possible improvements associated with interleukin-6 blockade in several measures. It is important to consider that other pathways that contribute to disease activity could also be relevant targets of therapy.^17,24,25

Safety data were consistent with the known safety profile of sarilumab.²⁸ Neutropenia and diarrhea were more common in the sarilumab group, which was consistent with findings from the trials of sarilumab in patients with rheumatoid arthritis. In addition, patients in our trial had more arthralgia than was observed in patients with rheumatoid arthritis who were receiving sarilumab.²⁹

A key limitation of our trial is the premature termination because of the Covid-19 pandemic, which reduced the sample size and statistical power. The trial protocol allowed a reduction in the sarilumab dose from 200 mg to 150 mg in patients with abnormal laboratory values, which occurred in only 1 patient. Thus, no conclusions can be made on the efficacy or safety of sarilumab at a dose of 150 mg in patients with polymyalgia rheumatica. Finally, the safety database is limited because of the small sample size.

In patients with polymyalgia rheumatica who had a relapse while tapering prednisone therapy, those who received sarilumab plus a 14-week prednisone taper had a higher frequency of remission and lower glucocorticoid exposure than those who received placebo plus a 52-week prednisone taper.

Supported by Sanofi and Regeneron Pharmaceuticals. Disclosure forms provided by the authors are available with the full text of this article at NEJM.org. A data sharing statement provided by the authors is available with the full text of this article at NEJM.org. We thank the following contributors to the trial conception and design: Anita Boyapati (Regeneron), Miguel Angel Gonzalez- Gay (University of Cantabria), Sarah Mackie (University of Leeds), Fei Cao-Ghoul (Genethon), Peter Villiger (University of Bern), Peter Wung (Sanofi and now at AbbVie), Ray Tao (Sanofi and now at Pfizer), and Veronique Gollot (Sanofi Chilly Mazarin), in addition to Vanessa Marks (Sanofi Chilly Mazarin), Wolfgang Schmidt (Immanuel Krankenhaus Berlin-Buch), and Ying Liu (Sanofi), who also provided assistance with the acquisition, analysis, and interpretation of the data; and Kerri Ford, Stefano Fiore, and Hubert Van Hoogstraten (all of Sanofi), who contributed to the data analysis and interpretation. Scientific writing support was provided by Vijay Kadasi and Kavita Garg (Sanofi).