The Effect of Pulmonary Function Test Reference Equations on Classification of Restrictive Lung Disease Severity in Systemic Sclerosis

Abstract

Background:

Interstitial lung disease (ILD) is a significant cause of morbidity and mortality in systemic sclerosis (SSc), particularly among Black patients. Pulmonary function tests (PFTs) are critical to screen for and monitor SSc-ILD. We examined whether race-specific and race-neutral PFT reference equations impact classification of restrictive lung disease (RLD) severity in Black and White patients with SSc.

Methods:

Baseline percent predicted forced vital capacity (ppFVC) was calculated for self-identified Black (N=641) and White (N=2909) patients in the Johns Hopkins Scleroderma Center Research Registry using race-specific (Global Lung Initiative 2012 [GLI 2012], National Health and Nutrition Examination Survey III [NHANES]) and race-neutral (GLI Global) equations. The percentage of Black and White individuals who switched RLD severity categories (normal (ppFVC≥80%); mild (70≤ppFVC<80%), moderate (60%≤ppFVC<70%), severe (50%≤ppFVC<60%) or very severe (ppFVC<50%)) when using race-neutral versus race-specific equations was calculated. The percentage who would meet typical ppFVC thresholds for immunosuppression, clinical trial eligibility, and lung transplant referral was compared.

Results:

Black individuals had lower absolute FVC values than White individuals. 47% (n=303) of Black individuals were reclassified as having more severe RLD and 17% (n=487) of White individuals were reclassified as having less severe RLD when using the GLI Global versus GLI 2012 equations. Statistically greater proportions of Black individuals met ppFVC thresholds for immunosuppression, clinical trial eligibility and lung transplant referral with race-neutral versus race-specific equations.

Conclusions:

The use of race-specific PFT reference equations may result in systematic misclassification of ILD severity with potential impact on healthcare delivery and clinical trial eligibility.

Introduction:

Systemic sclerosis (SSc) is a complex autoimmune disease characterized by fibrosis, vasculopathy and immune dysregulation. Individuals with self-identified Black race have higher incidence of SSc and develop SSc at a younger age than their White counterparts.¹ Furthermore, Black individuals with SSc face higher rates of interstitial lung disease (ILD), severe restrictive lung disease and overall mortality as compared to White individuals with SSc.^1–3 Though the etiology of these differences is not well understood, previous studies have suggested that it is multifactorial, with differences in genetic background, immune responses, and socioeconomic status likely playing a role. ^4–10

Examining the tools we use to diagnose and monitor SSc-ILD is important in elucidating these differences in health outcomes. Pulmonary function tests (PFTs), particularly assessment of the percent predicted forced vital capacity (ppFVC), are used for screening and monitoring of SSc-ILD. The American College of Rheumatology Guidelines for Screening and Monitoring of ILD in People with Systemic Autoimmune Rheumatic Disease recommend that individuals with SSc should undergo regular screening with PFTs and individuals with diagnosed SSc-ILD should undergo monitoring with PFTs every 3–6 months at the beginning of their illness.¹¹ Thus, in clinical practice, individuals with SSc undergo regular monitoring with PFTs at least yearly, but often more frequently. The ppFVC is a key measure in identifying restrictive lung disease (RLD); a ppFVC of less than 80% or a decline in ppFVC of greater than 10% is highly concerning for RLD and thus ILD among individuals with SSc. As such, the accuracy of ppFVC is of critical importance in clinical decision-making in the context of diagnosis and management of ILD. Decisions about whether to initiate immunosuppressive or antifibrotic therapy, clinical trial eligibility and lung transplant referral factor in the ppFVC as a surrogate for ILD severity.

Historically, predicted values for FVC were calculated using equations such as the third National Health and Nutritional Examination Survey (NHANES III) or the Global Lung Function Initiative 2012 (GLI 2012), which included age, sex, race and height as inputs.^12–16 However, the use of race as a factor in PFT reference equations lacks a biological basis and thus has been questioned.^17,18 Previous studies suggest that race-specific equations may not be accurate predictors of lung pathology, and may result in underdiagnosis and undertreatment of disease in Black patients.^17,19–24 In this setting, a race-neutral set of equations known as the GLI Global equations have been developed.^25,26 The American Thoracic Society now recommends the use of these reference equations in characterization of PFTs.¹⁸

The impact of race-specific and race-neutral equations on RLD classification in Black and White patients with SSc has not been evaluated. Importantly, evidence evaluating risk factors for and management of SSc-ILD has been based on data using race-specific equations for calculation of ppFVC, highlighting the need for reevaluation in the context of accurately defined outcomes. This study aims to examine the impact of race-specific (NHANES, GLI 2012) and race-neutral (GLI Global) reference equations on the classification of RLD and healthcare delivery for self-identified Black and White individuals with SSc via analysis of a large, single center SSc cohort.

Patients and Methods:

Study population:

Established in 1990, the Johns Hopkins Scleroderma Center Research Registry (JHSC) contains sociodemographic, clinical and serologic characteristics from a population of over 4400 individuals with SSc. This study analyzed routinely collected data and had local ethics committee approval. Adult patients (≥18 years of age) who had at least one PFT including spirometry measurements and self-identified as Black or White race were included. All individuals met the 1980 or 2013 American College of Rheumatology (ACR) classification criteria for SSc, had at least 3/5 features of CREST (calcinosis, Raynaud’s phenomenon, esophageal dysmotility, sclerodactyly and telangiectasia) syndrome or had definite Raynaud’s phenomenon, abnormal nailfold capillaries and a scleroderma-specific antibody.^27,28 PFTs were captured from testing at various sites including our academic medical center and local testing sites. We performed sensitivity analyses by using a restricted sample of patients who met the 2013 ACR classification criteria.

Scleroderma characteristics:

Clinical and demographic characteristics were obtained from the Registry. Participants self-reported race and sex, choosing from a fixed set of categories. Age of disease onset was defined by the age at the first non-Raynaud’s SSc symptom. Disease duration refers to the time between first non-Raynaud’s SSc symptom and time of first available PFT. History of smoking was a binary variable with individuals characterized as “ever” vs “never” smokers. Autoantibody testing was performed in patients who consented to provide a serum sample for research using the EuroBlotOne platform (Euroline Systemic Sclerosis Profile, Euroimmun Diagnostics, Lubeck, Germany). This assay tests for antibodies against centromere, RNA polymerase III (referred to as POLR3), Scl70, PM/Scl, NOR90, Th/To, Ku, U3RNP, Ro52 and PDGFR. Anti-centromere, anti-POLR3 and anti-PM/Scl testing included testing for two subunits each, and patients were defined as positive if both subunits were positive. Sera were assayed for anti-U1RNP antibodies using a commercial ELISA (Inova Diagnostics). Interstitial lung disease was defined by the presence of fibrosis on chest radiograph, computed tomography imaging of the chest, or presence of crackles on exam. Pulmonary hypertension (PH) was defined by a right ventricular systolic pressure of greater than 45 mmHg on echocardiogram.²⁹ Criteria for myopathy included abnormal muscle enzymes, myopathic EMG, abnormal muscle MRI or abnormal muscle biopsy. Clinical characteristics were considered present if criteria were met at any time during follow up. To test whether the distributions of variables are similar in Black and White patients, we used two sample t-tests assuming unequal variances for continuous variables and Chi-square tests for categorical variables. P-values are adjusted using the Benjamini-Hochberg method to control for the False Discovery Rate.³⁰

PFT Reference equations:

Three sets of reference equations were examined. In this section, we use terminology for racial groups as defined in the original papers for the development of NHANES III and GLI 2012. The NHANES III equations, developed in 1999, calculate predicted FVC values using separate reference equations for Caucasian, African American, and Mexican American adults and youths for each sex.¹² The predicted normal FVC values are calculated as a function of age and height. The GLI 2012 equations also calculate predicted FVC values based on age, height, sex, and race.¹³ Reference equations were derived for ethnic/racial categories defined as Caucasians, African Americans, and North East and South East Asians. In GLI 2012, log-transformed FVC is modeled as a function of log-transformed height and age, a smoothing age spline, and racial group as an indicator variable. Individuals who do not identify themselves with any of these four race groups or are of mixed ethnic origin are categorized into an “other” group, which serves as a reference for calculations of normal FVC. The GLI Global equation is a single race-neutral equation developed by using an inverse probability weight assigned to each observation of the GLI spirometry data based on race stratified by sex.²⁵ The general form of equation is identical to that of GLI 2012, but race is no longer used as an input for calculations. For each of the three approaches, a percent predicted FVC (ppFVC) value is obtained by dividing the measured FVC value by the predicted normal FVC value expressed as a percentage.

Percent predicted FVC values were calculated for Black and White individuals from the earliest PFT available in the JHSC registry using race-specific (NHANES III, GLI 2012) and race-neutral (GLI Global) equations. Age and height captured at the time of first-available PFT were used to calculate the predicted FVC. Self-identified race was used for the NHANES III and GLI 2012 equations.

Based on the American Thoracic Society recommendations for interpretation of spirometry, we additionally converted raw FVC values to z-scores based on NHANES III, GLI 2012, and GLI Global equations.¹⁵ Raw FVC values, age, height, and sex were used to calculate z-scores using GLI Global equations. Z-scores using NHANES III and GLI 2012 equations additionally required ethnicity or race.

Percent predicted DLCO (pDLCO) was calculated using GLI 2017 equations, as these are the latest equations available.³¹ Notably, these reference equations were only available for White individuals.

Outcome measures:

To study the general impact of using race-specific and race-neutral equations on different age, sex, and race groups, we plotted ppFVC by age and sex using the NHANES III, GLI 2012 and GLI Global equations. The average raw FVC, height, and age in ten-year age brackets were used to calculate the groupwise ppFVC. This approach allows us to visualize the difference in the ppFVC that is entirely attributable to using race-specific equations compared to using race-neutral equations for an average person with the same FVC, height, sex, and age in each age bracket.

For each patient, baseline lung function was classified as being normal (≥80% ppFVC) or as being indicative of mild (70–79% ppFVC), moderate (60–69% ppFVC), severe (50–59% ppFVC) or very severe (<50% ppFVC) RLD. The number and proportion of individuals who switched classification categories with the use of the race-neutral (GLI Global) and race-specific (GLI 2012 or NHANES III) equations were identified.

Using z-scores, lung function was categorized into normal (z-score >−1.645) or mild (−2.5 ≤ z-score ≤−1.645), moderate (−4 ≤ z-score < −2.5), or severe (z-score < −4) RLD.¹⁵ We used Wilcoxon signed rank tests to test whether there is a non-zero difference between lung function classification when using the race-neutral equations compared to using race-specific equations for all patients and separately for each race. Contingency tables of RLD categories defined by race-neutral and race-specific equations were generated. We used clinically relevant ppFVC thresholds for immunosuppressive medication initiation, clinical trial eligibility and referral to lung transplant as a theoretical exercise to assess the potential impact of different PFT reference equations on medical decision-making. The ppFVC thresholds for clinical trial eligibility were based on the inclusion criteria for the Scleroderma Lung Study (SLS) I and SLS II; these studies included individuals with a ppFVC between 45 and 79%.^32,33 In clinical practice, the decision to initiate immunosuppressive therapy is often determined by varying thresholds of RLD or a significant decline in ppFVC (≥10%); in our study, we used a conservative threshold of having at least moderate RLD, defined by ppFVC<70%.^34,35 We used a conservative threshold of ppFVC≤50% to identify patients who would be candidates for lung transplant referral. This threshold was based on the average ppFVC of SSc-ILD patients who underwent lung transplantation in a European multi-center cohort study.³⁶ For each race, McNemar tests were used to compare the proportion of individuals who met each clinical criteria using race-neutral and race-specific equations.

Results:

This study included 2909 self-identified White and 641 self-identified Black individuals from the Johns Hopkins Scleroderma Center. Demographic characteristics of the two groups are shown in Table 1. Black individuals had an earlier age of SSc onset (41.0 vs 43.7 years, p<0.0001), shorter disease duration (5.4 vs 6.6 years, p=0.0005) and younger age at the time of spirometry (48.3 vs 54.4 years, p<0.0001) as compared to White individuals. Diffuse cutaneous SSc was more common among Black individuals while limited cutaneous SSc was most common among White individuals. The autoantibody distributions of the two groups were distinct; anti-centromere, anti-RNA polymerase III, anti-PM/Scl were enriched among White individuals, and anti-U1RNP, anti-fibrillarin, anti-Scl70, anti-Ku, and anti-Ro52 were more common among Black individuals. Both groups were predominantly female. Black individuals had more severe lung disease on initial PFTs as compared to White individuals with lower absolute value of FVC (2.4 L vs 3.0 L, p<0.0001) and total lung capacity (TLC: 3.8 L vs 4.8 L, p<0.0001). Black individuals had lower baseline diffusing capacity of carbon monoxide (DLCO) (12.9 mL/min/mmHg vs 16.0 mL/min/mmHg, p<0.0001) and percent predicted DLCO (pDLCO) as calculated by the GLI 2017 equations (57.4% vs 73.9%, p<0.0001). In addition, echocardiographic evidence of pulmonary hypertension (48.5% vs 33.5%, p<0.0001) and myopathy (30.7% vs 14.5%, p<0.0001) were more common during the disease course in Black patients than White patients, while a history of smoking was less common (38.3% vs 48.5%, p<0.0001).

Table 1:

Demographics of Black and White individuals with SSc from the Johns Hopkins Scleroderma Center Research Registry.

Variable	White (mean (SD) / N (%)) n=2909	Black (mean (SD) / N (%)) n=641	P-value
Age (at time of spirometry)	54.4 (13.3)	48.3 (13.8)	<0.0001
Age of onset (years)	43.7 (15.1), N = 2,772	41.0 (14.2), N = 598	<0.0001
Sex			0.8759
Female	2,404 (82.6),	527 (82.2),
Male	505 (17.4)	114 (17.8)
Smoking (ever)	1,385 (48.5), N = 2,856	240 (38.3), N = 626	<0.0001
Disease duration (years)	6.6 (8.3), N = 2,852	5.4 (7.2), N = 633	0.0005
Cutaneous type			<0.0001
Sine	204 (7.0), N = 2,906	37 (5.8)
Limited	1,723 (59.3)	250 (39.0)
Diffuse	979 (33.7)	354 (55.2)
Height (m)	1.65 (0.09)	1.66 (0.09)	0.0195
FEV1 (L)	2.4 (0.7), N = 2,903	1.9 (0.6), N = 640	<0.0001
FVC (L)	3.0 (0.9)	2.4 (0.7)	<0.0001
DLCO (mL/min/mmHg)	16.0 (6.0), N = 2,668	12.9 (5.8), N = 588	<0.0001
ppDLCO*	73.9 (24.2), N = 2,668	57.4 (23.9), N = 588	<0.0001
TLC (L)	4.8 (5.4), N = 2,509	3.8 (1.2), N = 528	<0.0001
ILD	1,082 (40.2), N = 2,691	362 (60.8), N = 595	<0.0001
PH	737 (33.5), N = 2,198	234 (48.5), N = 482	<0.0001
Myopathy	409 (14.5), N = 2,818	189 (30.7), N = 616	<0.0001
Scleroderma renal crisis	120 (4.1)	34 (5.3)	0.2517
Antibody
Centromere	867 (32.0), N = 2,709	53 (8.9), N = 595	<0.0001
Scl70	453 (19.2), N = 2,354	154 (29.4), N = 523	<0.0001
POLR3	428 (19.1), N = 2,240	59 (12.1), N = 489	0.0004
Fibrillarin	84 (4.0), N = 2,118	91 (20.2), N = 450	<0.0001
Ku	69 (3.3), N = 2,118	47 (10.4), N = 450	<0.0001
NOR90	85 (4.0), N = 2,118	20 (4.4), N = 450	0.8374
PM/Scl	64 (3.0), N = 2,118	2 (0.4), N = 450	0.0038
Ro52	564 (26.6), N = 2,118	144 (32), N = 450	0.0283
Th/To	184 (8.7), N = 2,118	39 (8.7), N = 450	1.0000
U1RNP	171 (9.2), N = 1,853	111 (27.1), N = 410	<0.0001

Disease onset is defined by first non-Raynaud’s symptom. Disease duration is defined as years since onset to first PFT. FEV1: Forced expiratory vital capacity in one second; FVC: Forced vital capacity; TLC: total lung capacity; POLR3: RNA Polymerase III Ab; Scl70: anti-topoisomerase Ab. Interstitial lung disease (ILD) was defined by the presence of fibrosis on chest radiograph, computed tomography imaging of the chest, or presence of crackles on exam. Pulmonary hypertension (PH) was defined by a right ventricular systolic pressure of greater than 45 mmHg on echocardiogram. Criteria for myopathy included abnormal muscle enzymes, myopathic EMG, abnormal muscle MRI or abnormal muscle biopsy. Two sample t-tests assuming unequal variances were used for continuous variables and Chi-square tests were used for categorical variables. The presented p-values are adjusted using the Benjamini-Hochberg method to control for the False Discovery Rate.³⁰

^*ppDLCO: percent predicted DLCO calculated using GLI 2017 formula at baseline

Figure 1 illustrates the differences in calculation of ppFVC by age decile using the NHANES III, GLI 2012 and GLI Global equations. When comparing two individuals of the same sex, height and absolute value FVC, the race-specific Black equations (NHANES III, GLI 2012) result in a higher ppFVC value as compared to the race-specific White equations. Among Black patients, the use of race-specific equations results in a higher calculated ppFVC across age groups compared to the race-neutral (GLI Global) equations for both male and female individuals; in contrast, race-specific equations result in lower calculated ppFVC across age groups and sex strata for White patients compared to race-neutral equations.

Figure 1: The effect of race-specific vs race-neutral equations on the calculation of ppFVC.

The average ppFVC was calculated in ten-year age brackets using race-specific (NHANES III, GLI 2012) and race-neutral (GLI Global) equations for male and female individuals with SSc. For each age bracket, ppFVC is calculated using the same average FVC and height based on either a race-neutral or race-specific equation. The solid lines represent calculation of ppFVC with the GLI 2012 and NHANES III equations for Black race. The small, dashed line represents ppFVC calculated using the GLI 2012 and NHANES III equations for White race. The green line with larger dashes represents ppFVC as calculated by GLI Global.

Figure 2 demonstrates how the use of race-specific versus race-neutral reference equations affects RLD classification for Black and White individuals. The shaded regions demonstrate the percentage of individuals who switched severity categories with the use of different equations. In total, 47% (n=303) of Black individuals were reclassified to more severe RLD categories when using the GLI Global vs the GLI 2012 equations. Fifteen percent of Black individuals with normal lung function based on ppFVC (ppFVC≥80%) as per GLI 2012 equations had mildly restricted lung volumes (ppFVC of 70–79%) when calculated with the GLI Global equations. In contrast, the use of the GLI Global equations resulted in a shift toward less severe RLD for White individuals as compared to the GLI 2012 equations. Seventeen percent (n=487) of White individuals were reclassified to less severe RLD categories with the use of the GLI Global as compared to the GLI 2012 equations. Eight percent of White individuals who were classified as having mild RLD (ppFVC of 70–79%) as per GLI 2012 equations had normal calculated ppFVC values when using the GLI Global equations. Similar trends were observed when comparing NHANES III and GLI Global equations (Supplemental Figure 1); 53% of Black individuals were classified as having more severe and 25% of White individuals were classified as having less severe RLD when using GLI Global versus NHANES III equations.

Figure 2: Changes in classification of RLD severity with GLI 2012 vs GLI Global equations by race.

The alluvial plots demonstrate the number and percentage of individuals who switch restrictive lung disease impairment categories defined by ppFVC when comparing the race-specific GLI 2012 reference equations and the race-neutral GLI Global equations. Individuals for whom the use of race-neutral equations resulted in more severe restrictive lung disease are depicted by red alluvia. Individuals for whom the use of race-neutral equations resulted in less severe restrictive lung disease are depicted by green alluvia. The white bars represent patients who remain in the same restrictive lung disease category. The height of alluvia indicates the proportion of individuals switching categories when GLI Global equations are used.

Table 2 demonstrates how the use of race-specific versus race-neutral equations affects the number of individuals who met our previously defined thresholds for clinical trial eligibility, initiation of immunosuppression and referral for lung transplant. When comparing the GLI Global equations to NHANES III and GLI 2012, an increased proportion of Black individuals met commonly used ppFVC thresholds for clinical trial eligibility (69% vs 54% [p<0.0001] and 69% vs 57%, [p<0.0001]), initiation of immunosuppression (56% vs 39% [p<0.0001] and 56% vs 40% [p<0.0001]), and referral for lung transplant (14% vs 8% [p<0.0001] and 14% vs 9% [p<0.0001]). Notably, fewer White individuals met these clinical thresholds with the use of the GLI Global equations as compared to the NHANES III or GLI 2012 equations. There were no qualitative changes in the findings when we restricted the analysis to patients who met the 2013 ACR classification criteria (data not shown).

Table 2:

The Potential Effect of Race-Specific vs Race-Neutral PFT Reference Equations on Care Delivery.

	Eligibility for clinical trials enrollment N (%)			Initiation of immunosuppression N (%)			Referral for lung transplant N (%)
Equation	NHANES III	GLI 2012	GLI Global	NHANES III	GLI 2012	GLI Global	NHANES III	GLI 2012	GLI Global
All (N=3,550)	1,449 (41)	1,367 (39)	1,226 (35)	908 (26)	844 (24)	805 (23)	168 (5)	161 (5)	160 (5)
White (N=2,909)	1,100 (38)	1,000 (34)	783 (27)	660 (23)	585 (20)	449 (15)	115 (4)	102 (4)	69 (2)
Black (N=641)	349 (54)	367 (57)	443 (69)	248 (39)	259 (40)	356 (56)	53 (8)	59 (9)	91 (14)

This table demonstrates the numbers and percentages of individuals who meet three clinical criteria with each of the equations (NHANES III, GLI 2012, and GLI Global). Eligibility for clinical trials (SLS I/SLS II) is defined by ppFVC<80% and ppFVC≥45%, initiation of immunosuppression by ppFVC<70%, and referral for lung transplant by ppFVC≤50%. All p-values from McNemar tests comparing the proportions of individuals who meet each clinical criteria for each race when using race-specific (NHANES III and GLI 2012) and race-neutral (GLI Global) equations were <0.0001.

Similar results are observed when z-scores were used for lung classification. Supplemental Table 1 shows the distributions of individuals in each lung classification category (normal or mild, moderate, severe RLD) as defined by z-score with race-specific vs race-neutral equations. All four Wilcoxon signed rank tests that compared GLI Global equations to each of the GLI 2012 and NHANES III equations within White and Black individuals were statistically significant (p<0.0001). When the sample was restricted to those who met the 2013 ACR classification criteria, the results were unchanged (data not shown). These data confirm that using race-neutral equations compared to race-specific equations results in significant changes in lung function classification for both races. Supplemental Table 2 shows contingency tables of the same lung classification categories for each race.

Discussion:

In this study, we examined whether differing PFT reference equations could impact the classification of restrictive lung disease in Black and White patients with SSc and influence medical decision-making. Specifically, we evaluated the effect of race-neutral (GLI Global) and race-specific (NHANES III, GLI 2012) equations on the classification of RLD as defined by ppFVC among individuals with SSc through analysis of the Johns Hopkins Scleroderma Center Research Registry. Our analysis demonstrates that the use of race-specific PFT reference equations leads to clinically significant differences in classification of lung disease among Black and White individuals with SSc. When using the GLI Global equations, 47% of Black individuals were reclassified to more severe lung disease categories as compared to the GLI 2012 equations. In addition, significantly fewer White individuals would meet clinical ppFVC thresholds for immunosuppression initiation, referral for lung transplant, and clinical trial eligibility with the use of race-specific versus race-neutral equations. These data suggest that the historical PFT reference equations may have been a contributing factor to gaps in health outcomes amongst patients with SSc. Educating clinicians about the variation in reference equations over time, including the transition from race-specific to race-neutral equations, is important to ensure that these tests are being interpreted appropriately, both in a research context and when assessing an individual patient’s trajectory over time for medical decision-making.

SSc-ILD is a significant cause of morbidity and mortality for Black individuals with SSc. African ancestry is also associated with early pulmonary involvement, and increased burden of lung disease.^2,3 Diffuse skin disease, older age of disease onset, shorter disease duration and anti- topoisomerase 1 antibodies have been identified as risk factors for development or progression of ILD.^3,37 Many studies have attributed the increased rates of ILD among Black individuals to the high prevalence of anti-topoisomerase 1 antibodies. However, other studies suggest that the increased pulmonary disease seen among Black individuals with SSc is multifactorial in nature. While individuals of all backgrounds may be affected by evolving PFT reference equations, understanding the role measurement tools play in diagnosis, monitoring, and treatment decisions is particularly important in Black patients with SSc given their worse health outcomes.

The incorporation of GLI Global reference equations in the interpretation of pulmonary function tests will likely have significant clinical impact for the care of all individuals with SSc. A large cross-sectional study that analyzed all PFTs from an academic center PFT lab regardless of underlying disease process found an increased number of Black individuals with restrictive lung disease, as well as increased severity of the identified impairments with the use of race-neutral versus race-specific equations.²⁶ While our analysis largely focused on the effects of spirometry on Black and White individuals, the implementation of the GLI Global equations will impact individuals of other races as well. A large multi-center study comparing the GLI Global and GLI 2012 equations found that the newer equation resulted in the increase in prevalence of restrictive lung disease by 51% in Black individuals and 37.1% in Southeast Asian groups.³⁸

Further research is needed to understand the impact of varying PFT reference equations on medical decision-making across the broad spectrum of pulmonary diseases. A previous study that evaluated the impact of race-neutral equations on surgeon’s decision-making regarding lobectomy for Black individuals with lung cancer found that using race-neutral equations may result in less frequent recommendations for curative lung cancer surgery for Black individuals.³⁹ This study emphasizes the importance of a multimodal and holistic approach to decision-making for individuals with lung disease. While the use of race-neutral equations is now recommended by national associations such as the American Thoracic Society, debate exists in the field with regards to how to best implement these equations and it is not clear if race-neutral equations have been universally adopted. A recent longitudinal multicohort study found that race-neutral equations perform similarly to race-specific equations in predicting pulmonary outcomes but affect classification of lung disease. This same study demonstrated that the use of race-neutral equations has significant ramifications on other services such as occupational eligibility, disability payments and lung transplant priority.²⁴

For rheumatologists, awareness surrounding how pulmonary function test reference equations are derived and how these equations evolve over time is necessary in order to allow for effective clinical decision-making. For example, the recent transition to GLI Global equations may have resulted in shifts in ppFVC for White and Black individuals that are due to changing equations as opposed to actual clinical change. Without this important context, clinicians may make incorrect interpretations of a patient’s disease progression or trajectory. As such, education for rheumatologists with regards to these reference equations is essential.

This study evaluated the effect of race-neutral versus a race-specific equations on the classification of RLD as defined by ppFVC in a large cohort of Black and White individuals with SSc. Previous studies focused on Black individuals with SSc-ILD have been limited by small sample sizes; by using the Johns Hopkins Scleroderma Center Cohort, we were able to compare the use of these equations in a relatively large sample of SSc patients. However, our study had several limitations. Pulmonary function tests included in this study were obtained from multiple sites and were not limited to our single center. Given changing practice patterns over more than 30 years, all of our patients did not have high resolution chest CT or consistent use of right heart catheterization testing to define ILD or PH. Our analysis focuses on individuals with self-identified Black and White races but does not speak to the effect of race-specific or race-neutral equations on individuals of other races, or with mixed ethnicities. The American Thoracic Society/European Thoracic Society now recommends using z-scores to classify restrictive lung disease severity.¹⁵ While we included additional analysis related to RLD defined by z-scores, we framed our analysis on ppFVC as these values are used practically for clinical-decision making. As an exercise to demonstrate how the use of different PFT reference equations could potentially affect care delivery, we used conservative ppFVC thresholds for initiation of immunosuppression and referral for lung transplant recognizing the multifactorial nature of these decisions including disease trajectory over time. However, this study does not explicitly assess how the use of race-specific versus race-neutral equations affects clinical outcomes for Black and White individuals with SSc.

This study elucidates the impact of race-specific and race-neutral PFT reference equations on classification of RLD among Black and White individuals with SSc. More research is needed to understand how these, and other, measurement tools may impact the care that individuals with SSc and other systemic autoimmune rheumatic diseases receive. Changing reference equations over time can significantly affect pulmonary function test results and interpretation; further education with regards to how to appropriately use these tests moving forward is needed.

Significance and Innovations:

• This study demonstrates that the use of race-specific (GLI 2012; NHANES) versus race-neutral (GLI Global) pulmonary function test reference equations leads to clinically significant differences in classification of restrictive lung disease severity among Black and White individuals with systemic sclerosis.

• This study shows that more Black individuals with SSc would meet clinical forced vital capacity percent predicted (ppFVC) thresholds for immunosuppression initiation, lung transplant referral and clinical trial eligibility with the use of race-neutral versus race-specific equations.

• Rheumatologists should be aware of the ways in which PFT reference equations may evolve over time, as transitioning between different equations may result in shifts in ppFVC that are not reflective of true clinical change.