The diagnostic utility of rheumatoid factor and anticitrullinated protein antibody for rheumatoid arthritis in the Indian population

Abstract

Background

There is lack of Indian data on diagnostic utility of rheumatoid factor (RF) and anti-citrullinated protein antibody (ACPA) for diagnosis of rheumatoid arthritis (RA) and prevalence of these antibodies in patients with RA and the healthy population. The study was aimed to assess the diagnostic utility and prevalence of RF and ACPA at different titers in the Indian scenario.

Method

All the patients of RA fulfilling the European League Against Rheumatism/American College of Rheumatology (EULAR/ACR) 2010 classification criteria and age and gender-matched healthy controls were included in the study. RF and ACPA were measured by nephelometry and the enzyme-linked immunosorbent assay (ELISA) method, respectively.

Result

Of 803 patients (291 men and 512 women) included, the RF was positive in 566 (70.5%) study patients. The ACPA was positive in 527 (71.7%) patients of 735 of them. Among 408 healthy controls, 45 (11%) were RF positive and 19 (4.7%) were ACPA positive.

At the positive cutoff level, the RF had a specificity of 87.6% (95% confidence interval [CI] = 84.4–90.8; positive likelihood ratio [LR+] 5.7). Specificity at 2 and 3 times above the upper limit of normal (ULN) increased to 96.2% (95% CI = 94.3–98.1; LR+ 15.7) and 97.1% (95% CI = 95.5–98.7; LR+ 17.1), respectively.

The specificity of ACPA at the positive cutoff level was 94.4% (95% CI = 92.2–96.6; LR+ 12.7), which increased to 98% (95% CI = 96.6–99.4), at 2xULN level. The likelihood ratio for ACPA at all cutoff levels measured was more than 10.

Conclusion

The sensitivity and specificity of RF and ACPA in our study population are comparable with those of other studies. ACPA at lower titers may have sufficient diagnostic utility for RA in an appropriate clinical setting.

Introduction

Rheumatoid arthritis (RA) is the most common inflammatory arthritis encountered in clinical practice with worldwide prevalence of 0.5–1%.^1,2 Its prevalence in India has been reported to be 0.28–0.7%.³ Rheumatoid factor (RF) and anticitrullinated protein antibodies (ACPA) are serological hallmarks of the disease and have been an integral part of the American College of Rheumatology/European League Against Rheumatism (ACR/EULAR) classification criteria for RA.⁴ Sensitivity of RF and APCA has been found to vary in various studies. The RF has been reported to have an overall sensitivity of 60–80% and specificity of about 85%.^5,6 The ACPA has been found to be equally sensitive at about 67% but has significantly better specificity of about 98%.^7,6 The ACR/EULAR classification criteria define low-positive and high-positive titers for RF and ACPA with cutoff at 3 times the upper limit of normal (ULN) and ascribe higher weightage to the high titer subset.⁴ The available literature lacks the data on the prevalence of RF and ACPA in the healthy population, and there are no studies which have assessed the diagnostic utility of these antibodies in the Indian subset of patients with RA. The Indian population differs from the rest of the world in lifestyle, culture and genetic makeup, and the level of hygiene which generates curiosity whether the Indian population and subset of patients with RA follow the international trend with respect to the prevalence and diagnostic utility of RF and ACPA at various titers or not. In this study, we have assessed the sensitivity and specificity of RF and ACPA for RA at different titers and the prevalence of these antibodies in the healthy population.

Material and methods

This study was conducted at a tertiary care rheumatology center between March 2015 and March 2018. All the patients with RA fulfilling the ACR/EULAR 2010 classification criteria were included in the study irrespective of the disease duration.⁴ All patients underwent relevant clinical and laboratory evaluation to establish the diagnosis of RA. The RF was assessed by nephelometry in a Beckmen-Coulter nephelometer system, and the second-generation ACPA was measured by ELISA using the Aeskulisa RA/CP-Detect kit. The upper limits of normal for RF and ACPA were 20 IU/ml and 18 IU/ml, respectively. All patients were older than 16 years of age. In patients with only RF positivity, detailed clinical evaluation was performed to exclude other conditions associated with RF positivity if deemed necessary. Age and gender-matched healthy controls were included in the study. All the controls underwent thorough clinical evaluation to exclude any features of inflammatory arthritis and other features of connective tissue disease. Consent was obtained from each participant, and ethical clearance was taken from the institutional ethical committee. Sensitivity, specificity, and likelihood ratio for RF and ACPA were determined at the positive cutoff titer and incremental titers of >1.5xULN, >2xULN, >2.5xULN, and >3xULN. High- and low-positive titers for RF and ACPA were defined as per the ACR/EULAR 2010 classification criteria.⁴

Statistical analysis

Qualitative variables were expressed as numbers and percentages. Results for variables with normal distribution were expressed by their mean value and standard deviation (SD). For the comparison of normally distributed quantitative variables, parametric (t-test) tests were used, and the Chi-square test was used to compare the categorical variables. Receiver operating characteristic (ROC) analysis was used for the estimation of sensitivity and specificity at various titers, and positive likelihood ratio (LR+) was derived from the data. Statistical analysis was performed using SPSS, version 23.

Results

A total of 803 patients were included in the study. The value of RF was available for all 803 patients, whereas the ACPA value was available for only 735 patients. Thus, 735 patients had both RF and ACPA values. Among 803 patients, 291 were men and 512 were women (male:female ratio 1:1.7). The mean age of the patients was 44.09 (SD = 12.75) years, and they had RA for the mean duration of 75.72 months (SD = 74.02). A total of 408 healthy controls were included in the study. The demographic profile and RF and ACPA status of patients and controls are summarized in Table 1.

Table 1.

Demographic profile and RF and ACPA status in cases and controls.

Patients (N 803 for RF; 735 for ACPA)		Controls (N 408)
			Remarks
Age in years (mean, SD)	44.09 (12.75)	44.82 (13.13)	P 0.896
Male:female ratio	1:1.7 (291:512)	1:1.6 (160:245)	P 0.8
Disease duration in months (mean, SD)	75.72 (74.02)
RF in study and control groups
	N (%) M/F	N (%) M/F
RF negative	237 (29.5%) 113/124	363 (89%) 148/215
Positive RF	566 (70.5%) 178/388	45 (11%) 12/33
Positive RF more than 1.5XULN	492 (61.3%) 163/329	19 (4.7%) 5/14
Positive RF more than 2XULN	462 (57.5%) 157/305	14 (3.4%) 5/9
Positive RF more than 2.5XULN	429 (53.4%) 149/280	12 (2.9%) 4/8
Positive RF more than 3XULN	400 (49.8%) 149/251	11 (2.7%) 3/8
Low titer RF positive (<3XULN)	166 (20.7%) 29/137	34 (8.3%) 9/25
ACPA in study and control groups
ACPA negative	208 (28.3%) (74/134)	389 (95.3%) 155/234
Positive ACPA	527 (71.7%) 176/351	19 (4.7%) 5/14
Positive ACPA more than 1.5XULN	497 (67.6%) 170/327	10 (2.5%) 2/8
Positive ACPA more than 2XULN	469 (63.8%) 162/307	5 (1.2%) 1/4
Positive ACPA more than 2.5XULN	448 (45.9%) 156/292	4 (1.0%) 1/3
Positive ACPA more than 3XULN	398 (54%) 128/270	4 (1.0%) 1/3
Low titer ACPA positive (<3XULN)	129 (17.5%) 48/81	15 (3.7%) 4/11
RF and ACPA in study (N 735) and control groups (N 408)
Only RF positive	115 (15.6%)	37 (9%)
Only ACPA positive	122 (16.5%)	11 (2.7%)
RF and ACPA both positive	405 (55.1%)	8 (1.9%)
RF and ACPA both negative	93 (12.6%)	352 (86.2%)

ACPA: anticitrullinated peptide antibody; F: female; M: male; N: number; RF: rheumatoid factor; SD: standard deviation; ULN: upper limit of normal.

The RF was positive in 566 (70.5%) patientsof 803. Among these, 400 (49.8%) patients had a high-positive RF titer. In the control cohort, 45 (11%) were RF positive of which 11 (2.7%) had a high-positive RF titer.

Of 735 patients who had their ACPA value available, 527 (71.7%) patients were ACPA positive. A high-positive titer of ACPA was seen in 398 (54%) patients. Among controls, 19 (4.7%) were ACPA positive and only 4 (1%) controls had a high-positive ACPA titer.

Of 735 cases for which both RF and ACPA values were available, 405 (55.1%) cases revealed positivity for both the antibodies (double positive), while 93 (12.6%) cases were seronegative. The ACPA was present as a single positive antibody in 122 (16.6%) cases, while a single positive RF was seen in 115 (15.6%) cases. There were 214 cases of these 735 who did not express RF. Thus, the ACPA was positive in 57% (n 122) of RF-negative cases.

The sensitivity, specificity, and LR+ of RF and ACPA levels at different titers are given in Table 2. The ROC curves for RF and ACPA are depicted in Fig. 1, Fig. 2, respectively. At the positive cutoff level for the RF (>20 IU/ml), it had a specificity of 87.6% (95% CI = 84.4–90.8), sensitivity of 70.5% (95% CI = 67.3–73.7), and LR+ of 5.7. Sensitivity at >2xULN and >3xULN titers decreased to 57.5% (95% CI = 53.6–60.4) and 49.7% (95% CI = 46.2–53.2), respectively, while specificity at corresponding levels increased to 96.2% (95% CI = 94.3–98.1) and 97.1% (95% CI = 95.5–98.7), respectively. The likelihood ratio for RF at >2xULN and >3xULN titers increased to 15.7 and 17.1, respectively.

Table 2.

Sensitivity, specificity, and positive likelihood ratio of RF and ACPA at different titers.

Antibody titer	Sensitivity	95% CI	Specificity	95%CI	Positive likelihood ratio
RF 1xULN (>20 IU/ml)	70.5	67.3–73.7	87.6	84.4–90.8	5.7
RF 1.5xULN (>30 IU/ml)	61.1	57.6–64.4	93.6	91.2–96.0	9.5
RF 2xULN (>40 IU/ml)	57.5	53.6–60.4	96.2	94.3–98.1	15
RF 2.5xULN (>50 IU/ml)	53.4	49.9–56.9	96.6	94.8–98.4	15.7
RF 3xULN (>60 IU/ml)	49.7	46.2–53.2	97.1	95.5–98.7	17.1
ACPA 1xULN (>18 IU/ml)	71.1	67.8–74.4	94.4	92.2–96.6	12.7
ACPA 1.5xULN (>27 IU/ml)	67.6	64.2–71.0	96.6	94.8–98.4	19.9
ACPA 2xULN (>36 IU/ml)	63.8	60.3–67.3	98	96.6–99.4	31.9
ACPA 2.5xULN (>45 IU/ml)	61	57.5–64.5	98.2	96.9–99.5	33.9
ACPA 3xULN (>54 IU/ml)	54.1	50.5–57.7	98.4	97.2–99.6	33.8

ACPA: anticitrullinated peptide antibody; CI: confidence interval; RF: rheumatoid factor; ULN: upper limit of normal.

Fig. 1.

ROC curve for RF (AUC 0.847; 95% CI = 0.826–0,868). AUC: area and curve; CI: confidence interval; RF: rheumatoid factor; ROC: receiver operating characteristic.

Fig. 2.

ROC curve for ACPA (AUC 0.888; 95% CI = 0.870–0.907) ACPA: anticitrullinated protein antibody; AUC: area and curve; CI: confidence interval; ROC: receiver operating characteristic.

The sensitivity, specificity, and likelihood ratio of ACPA at the positive cutoff level (>18 IU/ml) was 71.1% (95% CI = 67.8–74.4), 94.4% (95% CI = 92.2–96.6), and 12.7, respectively. The sensitivity of ACPA at >2xULN and >3xULN titers was 63.8% (95% CI = 60.3–67.3) and 54.1% (95% CI = 50.5–57.7), respectively. The respective specificity at corresponding incremental levels was 98% (95%CI = 96.6–99.4) and 98.4% (95% CI = 97.2–99.6), respectively. The likelihood ratio for ACPA at all cutoff levels was more than 10.

Discussion

The presence of RF and ACPA has been the hallmarks of RA, and in an appropriate clinical setting, their presence helps in diagnosing RA at an early stage. In our study, we have analyzed the diagnostic utility of these antibodies along with their prevalence in healthy controls.

The RF has been reported to be positive in a wide range of 5–15% healthy individuals in different studies.^{8, 9, 10} The prevalence of RF has been reported to increase with age, and in the elderly population aged 65 years and older, it reaches up to 10%.¹¹ In our study, the prevalence of RF at 11% has corroborated with the other reports.

The available literature prevalence of ACPA has been found to be less than 2% in healthy individuals.^12,13 ACPA can be detected in about 20% of RF-negative patients with RA.¹⁴ However, in our study, the ACPA has been found in higher proportion (4.7%) in healthy controls than expected, and 57% of RF-negative patients with RA had positive ACPA. Higher prevalence of ACPA in the healthy Indian population and RF-negative patients with RA requires to be confirmed in further studies.

Sensitivity of RF and ACPA

In the available literature, the RF has been found to have a sensitivity of about 60–80% in the established disease, while the ACPA is found in 70–90% of the patients with RA.^5,7 Findings in our study are corroborating with RF and ACPA having similar sensitivity of 70.5% and 71.7%, respectively. Representative data on the sensitivity of RF and ACPA in the Indian subset of patients with RA are lacking. Among the available studies, in one of the studies from India on the prevalence of RA conducted in Pune, 62% of the RA cases exhibited RF positivity while all patients were found to be ACPA positive.¹⁵ Data from few other studies conducted primarily to report comorbidities and/or extraarticular manifestations among patients with RA revealed a range of RF seropositive cases from 68% to 90%.^{16, 17, 18, 19, 20, 21, 22, 23, 24, 25} Studies reporting ACPA sensitivity are even fewer. In one study conducted in Delhi, 90% of patients with RA were found to be ACPA positive.²¹ In one study by Gupta et al. ²⁶, which included 63 patients with RA and 51 controls, the sensitivity of ACPA was found to be 85.71%.

Specificity of RF and ACPA

The specificity of RF has been inferior as a low-titer RF has been found in 10%–15% of healthy individuals as per the available literature.¹⁰ In our study, 8.3% of controls have been found to have RF in low titers. The specificity of RF increases with increasing titers. In a meta-analysis by Nishimura et al. ⁶, RF had an overall specificity of 85% (CI = 84%–86%). The likelihood ratio of RF in this meta-analysis was 3.86, while at cutoff value for positive, our study showed LR+ of 5.7% for RF. In the same meta-analysis, pooled specificity for ACPA was 95% (CI = 95–96%) with the likelihood ratio of 12.32.⁶ Specificity of ACPA in our study at the cutoff value for positive was 94.4% with likelihood ratio of 12.7. Thus, specificity of RF and ACPA in our study matches the data available in the worldwide literature. Indian studies on serological markers of RA are inadequate. In the study by Gupta et al. ²⁶, specificity for ACPA was found to be 90.19%.

Diagnostic utility of RF and ACPA at various titers

With increase in titers, the specificity of RF increases. In the study by Nell et al. ²⁷, specificity of RF at 20 IU/ml increased from 89% to 96% at 50 IU/ml. In the 2010 ACR/EULAR criteria for classification of RA, incremental scores have been assigned for higher titers of RF and ACPA (score 2 and 3 for low- and high-positive titers, respectively).⁴ In our study, specificity and LR + for RF have shown a significant rise from that at the cutoff value for positive to the titer of >2xULN (specificity from 87.6% to 96.2% and LR+ 5.7 to 15). However, the gain in specificity and LR+ from the RF titer at 2xULN to 3xULN is only modest (specificity 96.2%–97.1% and LR+ 15 to 17.1). Thus, it appears that the diagnostic value of RF in our patients reaches to a significant level at a titer >2xULN itself.

The ACPA is the most specific serological marker of RA. In the study by Nell et al. ²⁷, it was found to be 98% specific. In our study, the ACPA at a titer >2xULN has been found to have 98% specificity (likelihood ratio 31.9) for RA. Thus, from findings of our study, it appears that RF at 2xULN and a positive ACPA at any level provide a significant posttest probability of RA in an appropriate clinical setting.

Seronegative RA

In our study, 12.6% patients with RA have been found to be seronegative (both RF and ACPA negative). In different cohorts in reported studies worldwide, 20-30% of patients have been found to be seronegative.^28,29

Our study is unique in being the only study in our knowledge to have assessed the diagnostic utility of RF and ACPA at different titers. Higher prevalence of ACPA in the healthy population and in RF-negative Indian patients with RA in our study is a finding of clinical importance. However, confirmation of these findings in further studies is required before it can be extrapolated to the whole subset of Indian population.

Our study has certain shortcomings. Firstly, the cases included in our study have established RA with the mean disease duration of 75.72 (SD = 74.02) months. The RF and ACPA levels have been reported to decrease with treatment.³⁰ Secondly, a better case to control ratio would have further consolidated the findings of our study. Thirdly, RF and ACPA are also found in other conditions which have not been considered in this study. However, either the prevalence of RF and ACPA in non-RA conditions is low or these conditions have low prevalence in themselves. And, moreover, RF and ACPA are an integral part of validated classification criteria, which includes other conditions supporting the diagnosis of RA, and this drawback can be overcome.

Conclusion

The prevalence of RF in Indian healthy controls and patients with RA corroborates with other studies. However, ACPA prevalence among healthy controls and among patients with RA who do not exhibit RF has been found to be higher than expected, which needs to be confirmed in larger studies. The sensitivity and specificity of RF and ACPA in our study population are comparable with those of other studies at titers of positive cutoff and low- and high-positive cutoffs. We may also infer that ACPA at lower titers may have sufficient diagnostic utility for RA in an appropriate clinical setting.

Disclosure of competing interest

The authors have none to declare.