Association of the Systemic Lupus International Collaborating Clinics Frailty Index and Damage Accrual in Long Standing Systemic Lupus Erythematosus

Abstract

Objective:

To externally validate the Systemic Lupus International Collaborating Clinics Frailty Index (SLICC-FI) in a prevalent SLE cohort and to assess the ability of the SLICC-FI to predict organ damage accrual among individuals with longstanding SLE.

Methods:

This was a secondary analysis of data from the Study of Lupus Vascular and Bone Long-Term Endpoints (SOLVABLE) cohort, which consists of adult women from the Chicago Lupus Database who met the 1997 revised ACR classification criteria for SLE. There were 185 SLE patients enrolled, of which 149 patients were included in a 5-year follow-up analysis. SLICC-FI and SLICC/ACR Damage Index (SDI) scores were calculated at baseline and 5-year follow-up. Unadjusted and adjusted logistic regression models estimated the association of baseline SLICC-FI scores (per 0.05 increase) with damage accrual at 5-year follow-up.

Results:

At enrollment the mean (SD) age of the 149 patients was 43.30 (10.15) years, mean (SD) disease duration was 11.93 (8.46) years, and mean (SD) SDI score was 1.64 (1.83). At baseline, the mean (SD) SLICC-FI score was 0.18 (0.08) and 36% of participants were categorized as frail (SLICC-FI >0.21). In a model adjusted for age, race and disease duration, each 0.05-unit increase in baseline SLICC-FI score was associated with a 39% higher odds of subsequent damage accrual (OR=1.39, 95% CI 1.05, 1.85).

Conclusion:

In a prevalent cohort of women with established SLE, higher baseline SLICC-FI scores were associated with higher risk of subsequent damage accrual at 5-year follow up.

Introduction:

Frailty is a concept long employed by geriatric medicine and more recently by other disciplines to assess an individual’s ability to respond to physiologic stress and quantify their susceptibility to adverse outcomes(1–3). Two primary approaches exist for measuring frailty in clinical practice. The frailty phenotype describes frailty as a clinical syndrome defined by the presence of certain clinical features including involuntary weight loss, exhaustion, slow gait speed, poor handgrip strength, and sedentary behavior(4). The frailty index (FI) operationalizes the concept through measurement of accumulated heterogeneous health deficits across multiple systems(4–6). The etiology of frailty is complex and includes chronic inflammation and immune activation as evidenced by elevated levels of interleukin (IL)-6, tumor necrosis factor (TNF) -α, and C-reactive protein(7, 8).

In systemic lupus erythematosus (SLE), the frailty phenotype has been associated with poor physical and cognitive functioning, functional decline, and increased mortality(9). Legge, et al. recently constructed the first lupus-specific frailty index, the Systemic Lupus International Collaborating Clinics Frailty Index (SLICC-FI), which demonstrated an association with increased mortality, damage accrual, and hospitalizations in the SLICC inception cohort(10–13). The SLICC-FI consists of 48 health deficits and integrates variables from traditional assessment domains (e.g. disease activity, organ damage, and health-related quality of life) into a single, comprehensive measure to predict outcomes in SLE(13).

The SLICC/American College of Rheumatology (ACR)-Damage Index (SDI), a well-validated and widely used instrument to measure accumulated damage in SLE, remains an important indicator of overall health and mortality risk for lupus patients(14). Preventing damage accrual is a primary focus of lupus treatment, although predicting which patients are at greatest risk for onset and progression of damage can be challenging. Given the significant variability in disease presentation, unpredictable course and heterogeneity of outcomes in SLE, the SLICC-FI could serve as a much-needed clinical tool to identify those patients most at risk for damage accrual and poor long-term outcomes.

While the SLICC-FI has been shown to predict adverse outcomes in a cohort of newly diagnosed SLE patients in which it was also derived, its generalizability to patients with established SLE remains unclear. The primary objective of this study was to apply the SLICC-FI to a cohort of SLE patients with longer disease duration and to assess the association of baseline SLICC-FI values with the risk of subsequent damage accrual. A secondary objective was to assess changes in SLICC-FI scores over time.

Materials and Methods:

Study Population:

This was a secondary analysis of data from the Study of Lupus Vascular and Bone Long-Term Endpoints (SOLVABLE) cohort which has been described in previous publications(15, 16). Briefly, SOLVABLE was a longitudinal epidemiologic study consisting of adult women from the Chicago Lupus Database (CLD) (age ≥18) who fulfilled the 1997 revised ACR classification criteria for SLE, as well as age-matched controls. All patients included in this study were invited to participate as part of the CLD for which informed consent was obtained for chart review to verify aspects of the medical record. The SOLVABLE study was approved by the Institutional Review Board at Northwestern University. A total of 185 SLE patients and 186 matched controls were enrolled. All participants provided informed consent. Patients were enrolled at an initial visit and reassessed at three- and five-year follow up between 2004–2013. This study used data from 149 SLE patients who participated in baseline and 5-year follow-up visits. Control patients were not included for this study as the SLICC-FI does not apply to non-SLE patients.

Data Collection:

At each study visit, patients completed the Medical Outcomes Survey Short Form 36 (SF-36), which is a validated measure of health-related quality of life, in addition to a self-administered questionnaire including demographic, behavioral, and medical history information. Trained physicians completed validated SLE-related instruments, including the Systemic Lupus Erythematosus Disease Activity Index 2000 (SLEDAI-2K) and the SDI. Additionally, all patients were evaluated by a trained physician, underwent physical examination, and completed blood and urine laboratory tests.

Adaptation of the SLICC-FI:

The SLICC-FI was calculated using methods previously described (13). Of the 48 health deficits included in the SLICC-FI, data for 46 deficits was collected in the SOLVABLE study. The two health deficits not included in the current analysis were anxiety (not collected in SOLVABLE) and myocarditis/endocarditis (excluded from SLICC-FI calculation due to >20% missing at baseline); each of these two health deficits had low prevalence in the original SLICC inception cohort (2.4% for anxiety and 1.4% for myocarditis/endocarditis) (13). In addition, the health deficit of Sjogren’s syndrome was not collected in the SOLVABLE cohort and was therefore obtained through chart review. In the original SLICC cohort, 100 iterations of the SLICC-FI were performed using 80% of the 48 total deficits selected at random and the descriptive statistics and distribution of SLICC-FI scores were largely unchanged. Therefore, it was reasonable to assume that applying the SLICC-FI with 46 health deficits should provide a stable and reliable measure of frailty(13).

For each of the 46 included health deficits, participants received a score between 0 and 1 (0=complete absence of deficit, 1=deficit is fully present). The sum of the scores from all available health deficits was then divided by the number of health deficits available for that patient. Patients were categorized as frail (SLICC-FI >0.21), least fit (0.10<SLICC-FI≤0.21) and relatively fit (SLICC-FI ≤0.10) at baseline using cut points derived from the general population and used in assessment of frailty for other rheumatic diseases (13, 17, 18). SLICC-FI scores were calculated for all patients at 5-year follow-up, with changes >0.03 considered clinically significant based on recommendations in the general population(19).

Measurement of Damage Accrual:

For each patient, the change in SDI was calculated by subtracting the baseline measure from 5-year follow-up measure. Damage accrual was defined as any increase in SDI score at 5-year follow-up.

Statistical Analysis:

Descriptive statistics were calculated to elucidate baseline demographic and clinical characteristics, including SDI scores. SLICC-FI scores were calculated for patients at the baseline and 5-year follow-up visits and the change in SLICC-FI score over time was reported.

Logistic regression models were used to evaluate the associations of baseline risk factors with subsequent 5-year damage accrual in this population. Results were reported as odds ratios (ORs) and 95% confidence intervals (CIs). Potential confounders associated with damage accrual in SLE were considered and univariable models were constructed for each potential confounder. A multivariable logistic regression model was fit to adjust for age, race, and disease duration, which are known risk factors associated with increased damage accrual, plus any confounding variables with p-values <0.1 in univariable screening (20, 21). Two sets of unadjusted and adjusted models for damage accrual were constructed to compare baseline SLICC-FI and baseline SDI scores as predictors of damage accrual.

Sensitivity analyses were performed by removing all overlapping SDI variables from the SLICC-FI, with the above models created for the remaining 32 SLICC-FI variables. To examine the influence of medication, additional models were analyzed controlling for corticosteroid, antimalarial, and conventional immunosuppressant use, which have each been shown to influence risk for future damage accrual(22, 23). All statistical analyses were performed using the SAS software version 9.4.

Results:

There were 149 patients with complete SLICC-FI data at both baseline and 5-year follow up. Of the 36 patients who did not have 5-year follow up visit, 20 declined participation, 3 did not respond, 2 were unable to be reached, 2 relocated, and 9 were deceased. Of the 9 deceased patients, cause of death was known in 5: 2 died from heart failure, 1 died from lung cancer, 1 died from cancer of unknown primary, and 1 died from infectious complications. The mean ± SD follow up time was 5.35 ± 0.6 years.

Baseline demographic and clinical characteristics of the dataset are shown in Table 1. These SLE patients had mean (SD) age of 43.30 (10.15) years and mean (SD) disease duration of 11.93 (8.46) years at baseline. The mean (SD) baseline SDI score was 1.64 (1.83) with median (IQR) of 1 (0–2). All baseline demographic and clinical characteristics between participants included and not included in this analysis were similar except current corticosteroid use (39.6% vs 58.3%, p=0.04).

Table 1:

Baseline demographic and clinical characteristics for patients in SOLVABLE cohort

Baseline Characteristics	Patients with 5-year follow-up, n=149	Patients without 5-year follow-up n=36	p value
Age (yrs), mean (SD)	43.30 (10.15)	41.31 (11.82)	0.308 a
Race			0.299 *b
White, n (%)	91 (61.07)	21 (58.33)
African American, n (%)	41 (27.52)	9 (25.00)
Hispanic, n (%)	8 (5.37)	5 (13.89)
Asian, n (%)	9 (6.04)	1 (2.78)
Education			0.598 *b
Less than high school diploma, n (%)	5 (3.36)	2 (5.56)
High school diploma, n (%)	44 (29.53)	14 (38.89)
College graduate, n (%)	61 (40.94)	13 (36.11)
Advanced degree, n (%)	39 (26.17)	7 (19.44)
Body Mass Index (kg/m2), mean (SD)	28.15 (7.51)	27.16 (7.15)	0.409 c
Disease duration (yrs), mean (SD)	11.93 (8.46)	12.13 (9.50)	0.908 c
SLEDAI-2K score, mean (SD)	3.85 (3.56)	5.17 (3.93)	0.052 c
SDI score, mean (SD)	1.64 (1.83)	2.06 (2.30)	0.410 c
Medication Use
Corticosteroids, n (%)	59 (39.60)	21 (58.33)	0.042 b
Hydroxychloroquine, n (%)	114 (76.51)	23 (63.89)	0.121 b
Immunosuppressants, n (%)	51 (34.23)	15 (41.67)	0.403 b

Note: SD=standard deviation, SLEDAI-2K= SLE Disease Activity Index 2000, SDI = Systemic Lupus International Collaborating Clinics Damage Index

^aT-test

^bChi-square test

^cWilcoxon Mann-Whiney test

^*P-value from testing of the overall effect for mutilevel categorical variable

The mean (SD) baseline SLICC-FI score was 0.18 (0.08) (Table 2). At baseline, 35.6% of participants were categorized as frail, 43.0% least fit, and 21.6% relatively fit (Figure 1). For the 36 patients without 5-year follow up, the mean (SD) baseline SLICC-FI score was 0.22 (0.10). Mean change (SD) in SLICC-FI at 5-year follow up was −0.01 (0.06), with 32 patients (21.5%) having a clinically significant increase in SLICC-FI score of >0.03, 46 patients (30.9%) having a decrease of >0.03 and 71 patients (47.7%) having no clinically significant change.

Table 2:

Change in frailty status from baseline visit to 5-year follow up in SOLVABLE cohort, n=149

	Baseline Visit	5-year follow up
SLICC-FI Mean (SD)	0.18 (0.08)	0.17 (0.09)
Distribution (min-max)	0.02 – 0.35	0.01 – 0.43
% Frail	35.57	30.20
% Least Fit	42.95	46.98
% Relatively Fit	21.58	22.82
Change in SLICC-FI, mean (SD)	-	−0.01 (0.06)
Change in SDI, mean (SD)	-	0.70 (1.30)

Note: SLICC = Systemic Lupus International Collaborating Clinics; FI = Frailty Index; DI=Damage Index; SD=Standard Deviation, IQR = Interquartile range

Figure 1:

Distribution of baseline SLICC-FI scores in SOLVABLE cohort

Note: Yellow line (0.10) represents cut off between relatively fit and least fit, red line (0.21) represents cut off between least fit and frail

At baseline, SLICC-FI scores and SDI scores had a positive correlation with Spearman’s rho r=0.33 (95% CI 0.18, 0.47). Baseline SLICC-FI score was not correlated with patient age. At 5-year follow up, 58 patients (38.9%) had an increase in SDI of ≥1, and 91 patients (61.1%) had no change in SDI. There were 33 patients (22.2%) with increase in SDI score of 1, 16 patients (10.7%) with increase of 2, and 9 patients (6.0%) with increase of 3 or more. In univariable analysis, none of the potential confounders were associated with the risk of damage accrual at 5 years (Table 3). In unadjusted analysis, each increase in baseline SLICC-FI score of 0.05-unit resulted in a 28% higher odds of subsequent damage accrual (OR 1.28, 95% CI 1.03–1.60). In multivariable analysis, baseline SLICC-FI remained statistically significant after controlling for age, race, and disease duration (adjusted OR 1.28, 95% CI 1.01–1.63) (Table 4). Baseline SDI score was not significantly associated with damage accrual at 5 years in this cohort (OR 1.02, 95% CI 0.85–1.22). When additional models were analyzed controlling for the effect of medication use at baseline, similar results were obtained for corticosteroid and anti-malarial use. However, when controlling for immunosuppressant use, the association between baseline SLICC-FI scores and 5-year damage accrual approached but did not meet statistical significance (Table 4). We performed sensitivity analyses removing all overlapping SDI items from the SLICC-FI and found similar results.

Table 3:

Univariate logistic regression models for the association between baseline demographic characteristics and damage accrual (baseline to 5 years), n=149

Baseline Independent Variable	Odds Ratio (95% CI)	p value
Age (yrs) at baseline	1.01 (0.98, 1.05)	0.416
Race/Ethnicity		0.105*
White	Referent
African American	0.97 (0.45, 2.08)
Hispanic	11.73 (1.38, 99.43)
Asian	0.48 (0.09, 2.44)
BMI	1.03 (0.98, 1.07)	0.261
Education		0.799*
Less than high school diploma	Referent
High school diploma	0.38 (0.06, 2.53)
College graduate	0.43 (0.07, 2.78)
Advanced degree	0.42 (0.06, 2.79)
SLE disease duration (yrs)	0.99 (0.95, 1.05)	0.618
SLEDAI-2K score	1.02 (0.93, 1.11)	0.756
SDI	1.02 (0.85, 1.22)	0.809
SLICC-FI cont (0.05 unit)	1.28 (1.03, 1.60)	0.024
Medication Use
Corticosteroids	0.89 (0.45, 1.75)	0.740
Plaquenil	0.94 (0.44, 2.05)	0.881
Immunosuppressants	1.48 (0.74, 2.95)	0.266

Note: SLICC = Systemic Lupus International Collaborating Clinics; FI = Frailty Index; DI=Damage Index, SDI = Systemic Lupus International Collaborating Clinics Damage Index

^*p-value from testing of the overall effect for multilevel categorical variable

Table 4:

Multivariable logistic regression models for the association of baseline SLICC-FI scores and baseline SDI scores with damage accrual (from baseline to 5 years) n=149

	Odds Ratio (95% CI)	p value
Model 1a: SLICC-FI (0.05-unit)	1.28 (1.01, 1.63)	0.04
Model 1–1b: SLICC-FI (0.05-unit)	1.30 (1.02, 1.67)	0.04
Model 1–2c: SLICC-FI (0.05-unit)	1.28 (1.01, 1.63)	0.04
Model 1–3d: SLICC-FI (0.05-unit)	1.26 (0.99, 1.60)	0.07
Model 2a: SDI (1-unit)	1.02 (0.83, 1.25)	0.86
Model 2–1b: SDI (1-unit)	1.02 (0.83, 1.26)	0.85
Model 2–2c: SDI (1-unit)	1.02 (0.83, 1.25)	0.85
Model 2–3d: SDI (1-unit)	1.00 (0.81, 1.23)	0.98

^a.Adjusted for age, race, and disease duration

^b.Adjusted for age, race, disease duration and corticosteroid use

^c.Adjusted for age, race, disease duration and antimalarial use

^d.Adjusted for age, race, disease duration and immunosuppressant use

Notes: SLICC = Systemic Lupus International Collaborating Clinics; FI = Frailty Index; SDI = Systemic Lupus International Collaborating Clinics Damage Index

Discussion:

This study demonstrates the SLICC-FI can be adapted for use in prevalent cohorts using pre-existing data that is commonly collected in SLE research studies. Our patients had a higher mean age of 42.9 years and longer mean disease duration of 11.97 years compared to the SLICC inception cohort, which had a mean age 35.7 years and median (IQR) disease duration 14.0 (10.7–18.4) months at baseline (13). The overall mean (SD) baseline SLICC-FI score was similar in our cohort at 0.18 (0.08) with a median (interquartile range) of 0.17 (0.10–0.23), compared to the SLICC inception cohort, which had a mean (SD) of 0.17 (0.08) and median (IQR) of 0.16 (0.11–0.22)(13). At baseline, 35.6% of participants were categorized as frail (SLICC-FI ≥0.21) compared to 27.2% of patients in the SLICC cohort(15). This discrepancy is expected given the differences in age and disease duration between the two cohorts.

In our cohort of patients with longstanding SLE, higher baseline SLICC-FI scores were associated with higher odds of damage accrual at 5-year follow-up, which maintained statistical significance after controlling for age, race, and disease duration. Additionally, controlling for baseline glucocorticoid use and antimalarial use produced similar results, whereas controlling for baseline conventional immunosuppressant use approached significance. The finding that frailty is a robust predictor of damage accrual in patients with longstanding SLE aligns with the concept that frailty arises via the accumulation of deficits within a complex network of interconnected elements, with greater deficit accumulation indicating a loss of physiologic reserve and increased susceptibility to future health threats (24). Frail individuals in general are more likely to accrue damage with new physiologic stressors (24).

In our study, baseline SDI scores did not predict subsequent damage accrual over 5 years. An abundance of evidence from cohorts with shorter disease duration has demonstrated that existing damage predicts risk of future damage(11, 20, 25, 26). However, a 2013 systematic review notes damage accrual appears to develop in a steady linear fashion over a number of years, with a potential plateau effect later in the disease course (21). Becker-Merok, et al found that damage clearly increased linearly over time up to 10 years, after which a flattening in the rate of damage accrual was observed (27). Our cohort had a mean (SD) disease duration of 11.93 (8.46) years at baseline, which may explain our finding that baseline damage may not be a sufficiently sensitive marker for subsequent damage accrual in patients with long-standing disease who may be accruing damage at a lower rate. Other studies have postulated this “plateau” phenomenon may be attributed to a healthy survivor effect, which could also be contributing in our population (28). The value of a frailty index lies in its ability to capture the complexity of human patients and the interaction between different health domains. It is the relationship between existing damage, ongoing disease activity, comorbidities, and patient quality of life that best captures the cumulative risk of future damage accrual in this population. The SLICC-FI may therefore be particularly useful in established cohorts for its ability to predict who is at highest risk for damage accrual during this plateau phase, though our study was not sufficiently powered test this hypothesis.

Importantly, frailty is reversible. Although on average frailty remained stable for the entire cohort over time, 31% of patients had a decrease in frailty at 5-year follow up. While the individual deficits contributing to these changes over time may vary from patient to patient, an advantage of the SLICC-FI is the integration of multiple deficits into a single predictor. Future studies investigating the trajectory of frailty in SLE patients over time are needed.

Our study has several strengths and limitations to note. The sample size was roughly 1/10^th the size of the SLICC cohort, resulting in decreased statistical power. Additionally, we enrolled only women who were primarily well-educated, which limits generalizability to both men and less well-educated groups. Importantly, future work should include the application of the SLICC-FI to patients more representative of the overall lupus population in the United States in terms of racial, socioeconomic, and geographic diversity. Finally, 36 patients did not have five-year follow-up visits, which may have introduced selection bias. However, apart from active corticosteroid use, the baseline characteristics of this group did not differ significantly from the group retained for 5-year follow-up analysis. This group did, however, have higher baseline SLICC-FI scores. Anticipating that patients with higher baseline frailty would experience more rapid damage accrual, it is unlikely that effect size was overestimated in this study. Given the low mortality rate in this cohort, the predictive validity of SLICC-FI values for mortality risk could not be evaluated, though this would be of interest in future work.

Overall, this study supports the usefulness of the SLICC-FI as a tool to predict damage accrual in SLE by demonstrating its generalizability to a patient population with longstanding disease. Unlike damage, frailty is reversible and future work should investigate how an individual’s SLICC-FI trajectory impacts health outcomes, as well as the potential usefulness of the SLICC-FI as an outcome measure in future SLE intervention studies.

Significance and Innovation:

• While the Systemic Lupus International Collaborating Clinics Frailty Index (SLICC-FI) has been shown to predict mortality, damage accrual, and hospitalizations in the SLICC inception cohort, this is the first study to externally validate the SLICC-FI in a prevalent cohort of women with more longstanding SLE.

• Baseline SLICC-FI scores were similar in this cohort of patients with longstanding SLE when compared to those with newly diagnosed disease in the SLICC inception cohort.

• The SLICC-FI predicted damage accrual in patients with longstanding SLE, whereas baseline SLICC/ACR Damage Index scores did not, reinforcing the usefulness of the SLICC-FI as a robust tool to identify patients at higher risk of adverse outcomes.

• Unlike damage, frailty is reversible and 30.9% of patients had clinically significant decreases in SLICC-FI scores at 5-year follow-up. Future studies should focus on understanding the impact of an individual’s frailty trajectory on health outcomes.