Neutrophil to Lymphocyte Ratio Predicts Glucocorticoid Resistance in Polymyalgia Rheumatica: A Prospective Study

Background Neutrophil to lymphocyte ratio (NLR) and platelet to lymphocyte ratio (PLR) have been found to correlate with disease activity in several rheumatic diseases, however their clinical and prognostic utility in polymyalgia rheumatica (PMR) remains unclear. This study evaluated the relationship between NLR and PLR, and disease activity and glucocorticoid resistance in PMR. Methods Data for disease activity (PMR-AS) and full blood examination was obtained from a prospective observational cohort comprising newly diagnosed (2012 EULAR/ACR classification criteria), steroid-naïve PMR patients treated with low-dose glucocorticoid therapy (BSR guideline). Glucocorticoid resistance was defined as non-response to initial prednisolone dose (15mg/day) or initial response followed by flare (PMR-AS ≥9.35 or ∆≥6.6) upon weaning prednisolone to 5mg/day. Univariable Bayesian linear regression analysis of the relationship between PMR-AS (baseline and mean) and NLR and PLR was performed. Predictors of glucocorticoid resistance were identified using a multivariable outcome model, with variables chosen based upon Bayesian model selection. Results Of the 32 included patients, 16 (50%) fulfilled the primary outcome measure of glucocorticoid resistance. These participants were older, typically female, and had higher baseline CRP than their glucocorticoid-responsive counterparts. A statistically significant relationship was identified between PMR-AS and both NLR (OR 28.1 [95% CI 1.6 – 54.7]) and PLR (OR 40.6 [95% CI 10.1 – 71.4]) at baseline, with PLR also found to correlate with disease activity during follow-up (OR 15.6 [95% CI 2.7 – 28.2]). Baseline NLR proved to be a statistically significant predictor of glucocorticoid-resistant PMR (OR 14.01 [95% CI 1.49 – 278.06]).


Background
Polymyalgia rheumatica (PMR) is an inflammatory condition characterised by subacuteonset shoulder and hip pain, and stiffness. Oral glucocorticoids represent the treatment mainstay and whilst cessation of therapy is the ultimate goal, up to 50% of patients still require prednisolone 2-3 years after diagnosis.(1) A higher baseline erythrocyte sedimentation rate (ESR) has been associated with disease relapse and lower probability of glucocorticoid discontinuation, however these findings have not been consistently reproduced. (2) To date, a reliable and readily accessible biomarker to measure disease activity and predict resistance to standardised low-dose glucocorticoid therapy has not been identified in PMR.
Neutrophil, platelet and lymphocyte counts undergo temporary changes in an active inflammatory state. Neutrophil to lymphocyte ratio (NLR) and platelet to lymphocyte ratio (PLR) can be calculated from a routine full blood examination (FBE) using the neutrophil count or platelet count divided by the lymphocyte count. In oncology, high NLR and PLR are associated with poor prognostic outcomes across a range of solid organ malignancies. (3,4) These indices have similarly been found to correlate with disease activity in several rheumatic diseases including most recently PMR, but their role in predicting treatment response in this condition is less well-defined. (5)(6)(7)(8) In this study, we sought to prospectively characterize the relationship between NLR and PLR, and disease activity in patients with newly diagnosed PMR. The ability of NLR and PLR to predict resistance to standardised low-dose glucocorticoid therapy was also evaluated.

Patients
Patients with newly diagnosed, steroid-naïve PMR according to the 2012 EULAR/ACR classification criteria (9)  Following written consent, demographic and clinical data were collected and the health assessment questionnaire-disability index (HAQ-DI) completed. A standard physical examination was undertaken including measurement of the shoulder range of motion (elevation of upper limb score). In order to measure disease activity using the PMRactivity score (PMR-AS)(10), participants marked a visual analogue scale indicating pain severity and the investigator provided a physician global assessment. Relevant differential diagnoses were excluded by testing creatine kinase, thyroid function, rheumatoid factor, anti-citrullinated peptide antibodies (ACPA) and anti-neutrophil cytoplasmic antibodies. FBE, C-reactive protein (CRP) and erythrocyte sedimentation rate (ESR) were performed at baseline and each subsequent visit. A whole body 18 F-Fluorodeoxyglucose ( 18 F-FDG) positron emission tomography/computed tomography (PET/CT) scan (Gemini-TF64 or Ingenuity-TF128 PET/CT system; Phillips, Cleveland, OH, USA) from skull vertex to toes was also undertaken in all participants within 7 days of study enrolment.

Treatment Protocol
Participants were treated with a standardised schedule of low-dose glucocorticoid therapy as outlined in the British Society of Rheumatology (BSR) guideline.(11) Follow-up visits were scheduled at weeks 4, 8, 16, 24, 32 and 46 (the minimum duration of this treatment protocol) and disease activity was re-calculated. A PMR-AS score of ≥ 9.35 in a patient who had achieved clinical remission (PMR-AS < 9.35) or increase in PMR-AS by ≥ 6.6 if the participant had been previously responding to treatment (PMR-AS ≥ 9.35 but falling between successive visits) constituted the definition of disease relapse. When this occurred, the patient's prednisolone dose was increased to the previous higher dose for 4 weeks and an additional study visit performed to determine if the weaning schedule could be reinstated. If relapse occurred on more than two occasions, methotrexate was commenced as a steroid-sparing agent, and the participant was withdrawn from the study.

Glucocorticoid Resistance
Glucocorticoid resistance has been previously defined by Mori et al. in PMR as nonresponse to an initial prednisolone dose of 15 mg/day or initial response followed by disease relapse upon weaning prednisolone to 5 mg/day. (12) This definition was accordingly utilised as the primary outcome measure for glucocorticoid-resistant disease in our study.

Statistical Analyses
All statistical analyses were conducted using R version 3.5.2 and the package rstanarm. (13,14) Results for continuous variables are reported as mean ± standard deviation or median (interquartile range). NLR and PLR, both right-skewed distributions, were logtransformed prior to analysis. Univariable Bayesian linear regression was used to analyse the relationship between NLR and PLR, and PMR-AS (baseline and mean). Several multivariable, Bayesian logistic regression models were compared using leave-one-out cross-validation -from the loo package(15) -to select predictors of resistance to low-dose glucocorticoid therapy. All outcomes are reported as odds ratios (OR) with 95% credible intervals (CI). (16) Weakly-informative prior distributions were applied using student tdistribution with 7 degrees of freedom. (17) Results Thirty-two of the 35 patients (91.4%) recruited at baseline were suitable for inclusion in the final analysis (Table 1): one patient was excluded due to large vessel vasculitis on whole body PET/CT consistent with concomitant GCA; another patient's final diagnosis was revised to Parkinson's disease; and the third patient required a modified schedule of glucocorticoid therapy due to a history of prednisolone-induced central serous retinopathy. Abbreviations: BMI -body mass index; EMS -early morning stiffness; CRP -C-reactive protein; ESR -erythrocyte sedimentation rate; PMR-AS -polymyalgia rheumatica-activity score; HAQ-DI -health assessment questionnaire-disability index.  There are several other limitations to this study. It involves a relatively small number of participants and consequently the results may not be generalizable to a larger, more

Availability of Data and Materials
The datasets used and analyzed during this study are available from the corresponding author on reasonable request.

Competing Interests
Author CO has received speaking honoraria from Roche, Janssen and Pfizer, and meeting