Risk Stratification in Transthyretin Cardiac Amyloidosis: The Added Value of Lung Spirometry

Transthyretin cardiac amyloidosis (ATTR-CA) is an increasingly recognized disease that often results in heart failure and death. Traditionally, biological staging systems are used to stratify disease severity. Reduced aerobic capacity has recently been described as useful in identifying higher risk of cardiovascular events and death. Assessment of lung volume via simple spirometry might also hold prognostic relevance. We aimed to assess the combined prognostic value of spirometry, cardiopulmonary exercise testing (CPET) and biomarker staging in ATTR-CA patients in a multi-parametric approach. We retrospectively reviewed patient records with pulmonary function and CPET testing. Patients were followed until study endpoint (MACE: composite of heart-failure-related hospitalization and all-cause death) or censure (1 April 2022). In total, 82 patients were enrolled. Median follow-up was 9 months with 31 (38%) MACE. Impaired peak VO2 and forced vital capacity (FVC) were independent predictors of MACE-free survival, with peak VO2 < 50% and FVC < 70% defining the highest risk group (HR 26, 95% CI: 5–142, mean survival: 15 months) compared to patients with the lowest risk (peak VO2 ≥ 50% and FVC ≥ 70%). Combined peak VO2, FVC and ATTR biomarker staging significantly improved MACE prediction by 35% compared to ATTR staging alone, with 67% patients reassigned a higher risk category (p < 0.01). In conclusion, combining functional and biological markers might synergistically improve risk stratification in ATTR-CA. Integrating simple, non-invasive and easily applicable CPET and spirometry in the routine management of ATTR-CA patients might prove useful for improved risk prediction, optimized monitoring and timely introduction of newer-generation therapies.


Introduction
Transthyretin cardiac amyloidosis (ATTR-CA) is an increasingly recognized condition which originates primarily from the abnormal extracellular accumulation of insoluble misfolded ATTR protein deposits within the myocardial interstitium [1]. Two types of ATTR amyloid fibrils are distinguished: non-mutated (ATTRwt) or variant transthyretin (ATTRv). 1,2 Progressive cardiac amyloid fibril accumulation is synonymous with restrictive heart wall chamber behavior and impaired myocardial contractile reserve, which often lead to symptomatic chronic heart failure and death [1,2].
Traditional clinical echocardiographic parameters and cardiac biomarkers have been accurately used to stratify disease severity in ATTR-CA patients [2][3][4]. Biological staging systems applicable to both wild-type and variant ATTR-CA, as well as echocardiographic parameters, have been described as having prognostic value [4]. Besides such well-established risk models, several authors have brought to light the pertinence of functional evaluation, such as peak aerobic capacity (pVO 2 ) and pulmonary function test, in identifying CA patients with poor prognosis [5][6][7][8][9][10]. The cardiopulmonary exercise test (CPET) is the gold standard test to determine prognosis in chronic heart failure patients with reduced ejection fraction. We and other groups have suggested that the combination of reduced peak VO 2 and NT-proBNP levels represents a valuable predictor of all-cause mortality and heart-failure-related hospitalization in ATTR-CA patients [5][6][7]10]. While being a safe and useful diagnostic tool in many pathological conditions, CPET requires sophisticated and expensive devices, as well as specialized medical staff, which may limit its clinical use, notably in low-resource settings. In contrast, spirometry is a simple method to evaluate pulmonary function, which is successfully used to stratify heart failure severity [11][12][13]. The use of spirometry to screen for ventilatory defects in patients with heart failure is known to improve risk stratification based on pVO 2 [13].
While restrictive spirometry pattern is a frequent observation in patients with ATTR-CA [8,9], the objective of the present study was to assess the potential link between the occurrence of major adverse cardiac events and impaired cardiopulmonary function (reduced pVO 2 and lung volume restriction) in ATTR-CA patients. We also analyzed the prognostic value of these two functional parameters in predicting adverse outcome, and quantified their added predictive ability when combined with a well-established validated prognostic score (ATTR biomarker staging). To the best of our knowledge, the combined prognostic value of these three markers has not yet been documented in ATTR-CA.

Study Context, Design and Population
This multicenter observational study involves four referral centers for cardiac amyloidosis (CA) management, namely the University Hospital of Martinique (Fort de France, France), University Hospital of Toulouse (Toulouse, France), the Fondazione Toscana G. Monasterio (Pisa, Italy) and the Mayo Clinic (Rochester, MN, USA). A retrospective review of the medical records of ATTR-CA patients was conducted.
Inclusion criteria were the conduct of pulmonary function (PFT) and cardiopulmonary exercise testing (CPET) whilst patients were in a stable condition at the four pre-cited expert centers from 1 August 2005 to 23 December 2021. A stable condition was defined as the absence of hospitalization for cardiac decompensation or any other cause in the six months prior to PFT and CPET. Patient exclusion criteria were the age below 18 years, pregnancy, breastfeeding, history of malignancy (other than AL amyloidosis), AL amyloidosis, serum amyloid A amyloidosis, chronic obstructive pulmonary disease, persistent asthma, positivity for human immunodeficiency virus, active hepatitis infection, pre-existing heart diseases and other causes of heart diseases (ischemic, hypertensive, valvular heart disease), or any other concurrent medical condition or disease that would likely interfere with study procedures or results.
The study was approved by the institutional review boards of the participating centers. All patients were managed in accordance with the amended Declaration of Helsinki (https:// www.wma.net/policies-post/wma-declaration-of-helsinki-ethical-principles-for-medical-researchinvolving-human-subjects) (accessed on 25 April 2023). Written informed consent from the patients or patients' legal guardian/next of kin was not required to participate in this study in accordance with the current legislation and the institutional requirements. Informed consent was obtained from all patients (University Hospital APHP IRB: 00006477, 1 May 2022).

Diagnosis of ATTR Amyloidosis
All participants underwent a thorough clinical, imaging and biological evaluation for ATTR amyloidosis diagnosis [2,14,15] including 12-lead electrocardiography and blood tests such as creatinine, cardiac troponin T (hs) and NT-proBNP. Echocardiography examination was performed in accordance with the recommendations from the American Society of Echocardiography. Cardiac involvement was evoked on ventricular hypertrophy, and a decrease in longitudinal global strain with abnormal apical texture was characterized as a speckled appearance. ATTR amyloidosis was diagnosed by cardiac uptake on the 99mTc-labeled phosphate bone scintigraphy in the absence of monoclonal gammopathy or abnormal free light chains in blood and urine. ATTR amyloidosis diagnosis was confirmed by histological demonstration of amyloid fibrils in salivary duct glands, subcutaneous adipose tissue or endomyocardial biopsies. Genetic testing for transthyretin mutation was performed in all ATTR-CA patients. Gillmore's validated three-stage biomarker staging score [15] was calculated for each patient according to the NT-proBNP level (cut-off 3000 ng/L) and estimated glomerular filtration rate (cut-off 45 mL/min/m 2 ). Stage I was defined as NT-proBNP ≤ 3000 ng/L and eGFR ≥ 45 mL/min, Stage III was defined as NT-proBNP > 3000 ng/L and eGFR < 45 mL/min, and the remainder of the patients were Stage II.

Pulmonary Function Testing (PFT)
PFT included spirometry, functional residual capacity and total lung volume, and lung diffusion capacity, which was performed on subjects in a sitting position using the guidelines of the European Respiratory Society (ERS) [16]. For study analysis, standard spirometry parameters were considered, which included absolute and percent-of-predicted normal FEV1 (forced expiratory volume in the first second) and FVC (forced vital capacity), as well as the FEV1/FVC ratio. Spiromery was considered as normal (FEV1/FVC ≥ 0.70 and FVC ≥ 80% of predicted values) or restricted (FEV1/FVC ≥ 0.70 and FVC < 80% predicted values) using race-based GLI (Global Lung Function Initiative) predicted values [17,18].

Cardiopulmonary Exercise Testing (CPET)
CPET was performed according to the standardized procedures using an upright electromagnetic braked cycle ergometer as recommended by the American Thoracic Society (ATS) guidelines [19]. Exercise testing contraindications included unstable cardiovascular diseases, orthopedic impairment compromising exercise performance and mental impairment leading to inability to cooperate.
The patients underwent a cardiopulmonary exercise test previously performed as "familiarization" in order to achieve suitable results. The test exercise protocol involved an initial 3 min rest period, followed by unloaded cycling for 2 min with a progressive 5-to 10-watt increment every minute until exhaustion at a pedaling frequency of 60-65 revolutions/minute (rpm). Such incremental phases lasting 8-12 min are efficient and provide suitable information, in particular at peak exercise. Subjects were continuously monitored using 12-lead ECG. Blood pressure was recorded every 2 min. Breath-by-breath cardiopulmonary data were measured at rest, warm-up and during incremental exercise testing. Subjects respired through an oro-nasal mask. Before each test, oxygen (O 2 ) and carbon dioxide (CO 2 ) analyzers and flow mass sensor were calibrated using the available precision gas mixture and a 3 L syringe, respectively. Minute ventilation (VE), oxygen uptake (VO 2 ) and carbon dioxide output (VCO 2 ) were recorded as concurrent 10 sec moving averages, which was the determined ventilation anaerobic threshold by the V-slope method.
Peak values were averaged over the last 30 sec of exercise. Peak oxygen pulse (O 2 pulse), a surrogate of stroke volume, was calculated and expressed in mL per beat and as a percentage of the predicted value by dividing the predicted peak VO 2 by the predicted peak heart rate (HR). Tidal volume and breathing frequency were measured online. Ventilatory reserve was calculated as (MVV-peak VE)/MVV * 100, where MVV is the maximal voluntary ventilation estimated as FEV1 multiplied by 35 [19]. Ventilatory efficiency, as indicated by the increment in VE relative to VCO 2 (VEVCO 2 slope), was calculated offline as a linear regression function using 10 s averaged values [19]. The participants were encouraged to continue on their exercise bout until a true symptom-limited exhaustive level was achieved. Patient effort was considered maximal if two of the following conditions were achieved: predicted maximal work, age-predicted maximal heart rate (HRmax), ventilatory O 2 equivalent VE/VO 2 > 45 and respiratory exchange ratio (RER, i.e., volume of carbon dioxide produced/volume of oxygen consumed > 1.10) [19]. Symptoms and subjective ratings of perceived exertion were recorded in order to estimate exertion level.

Follow-Up and Endpoints
The study's primary endpoint was the occurrence of a major adverse cardiac event (MACE) defined as the composite of heart-failure-related hospitalization or all-cause death. Patient follow-up was carried out from the time of spirometry/CPET until either MACE onset or censoring on 1 April 2022.

Statistical Analysis
Baseline patient, PFT and CPET characteristics were described and compared according to MACE occurrence during follow-up. For all descriptive and inferential analyzes, the assumption of normal data distribution was analyzed. Mean and standard deviations were reported for normally distributed variables and median and interquartile range (IQR) for non-normally distributed variables. Categorical variables were presented as absolute values and percentages. The following tests were used for group comparisons: Student t-test, Wilcoxon-Mann-Whitney test, Chi-square test and Fisher's exact test. Univariate and multivariate logistic regression models were first fitted to assess the independent effect of predictors on MACE occurrence. Receiver Operating Characteristic (ROC) Curve analysis was used to establish the optimal cut-off points for the outcome prediction by the relevant PFT and CPET variables. Time-to-event data were also evaluated with the use of Kaplan-Meier estimates and Cox proportional hazards methods. Due to the relatively small number of clinical events (MACE), the number of variables in multivariate regression models was restrained in accordance with the results of univariate logistic (p < 0.05) or Cox regression analysis (p < 0.15), as well as the clinical relevance of variables and collinearity. Variables fulfilling the latter conditions were retained for backward stepwise multivariate logistic or Cox regression analysis. The goodness-of-fit of final multivariate models was ascertained. In order to assess the added predictive ability of pertinent pulmonary function testing (PFT) and cardiopulmonary exercise testing (CPET) variables when combined with a wellestablished validated prognostic score (ATTR biomarker staging) 1, three metric measures were computed: Areas Under the ROC Curve (dichotomous and time-dependent analysis), Integrated Discrimination Improvement (IDI) and Net Reclassification Improvement (NRI). Areas Under the ROC Curve (AUC), IDI and NRI are complementary measures used to quantify improvement in a model's performance in predicting risk when new markers are added to the existing models [20][21][22]. AUC is a measure of model discrimination (i.e., how well the model separates the subjects who did and did not experience an event). It essentially depicts a tradeoff between the benefit of a model (true positive or sensitivity) vs. its costs (false positive or 1-specificity). AUC ranges from 0.5 (no discrimination) to 1 (perfect discrimination). While the difference in AUC is a common method to compare two models, it is relatively insensitive to detecting clinically important risk differences (1,2,3,7). IDI measures the new model's improvement in average sensitivity (true positive rate) without sacrificing its average specificity (true negative rate). In comparing the models, IDI measures the increment in the predicted probabilities for the subset experiencing an event and the decrement for the subset not experiencing an event. This adds an important element that AUC lacks. The AUC is a rank-based statistic in which all that matters is which probability is higher or lower, whereas IDI provides a measure of how far apart on average they are. NRI is based on a concept similar to IDI, but its focus is on the upward and downward movement of the predicted risks among those with and without events. NRI evaluates the "net" change in the proportion of subjects assigned a more appropriate risk or risk category (reclassification) using the new risk prediction model [20][21][22]. For study purposes, categorical NRI was computed according to clinically meaningful cut-points defining low, medium and high risk for MACE as follows: ≤0.47%, >0.47-0.70%, >0.7-0.93%, >0.93%. All statistical analyses were performed using the SAS software (version 9.4, Cary, NC, USA), with p-values < 0.05 considered as statistically significant.

Cardiopulmonary Functional Phenotype
Lung volume restriction, evidenced by a restrictive spirometry pattern as per GLI/ERS predicted values, was observed in 49% of ATTR-CA patients (61% in "MACE" patients vs. 41% in "no MACE" patients; p = 0.08). Patients with a restrictive spirometry pattern presented with a characteristic severe rapid shallow breathing pattern illustrated by a higher respiratory frequency.
The analysis of the main functional features issuing from the pulmonary function and the cardiopulmonary exercise testing highlighted reduced resting lung volumes, as well as an impaired aerobic capacity in the MACE patients (Table 2). Mean peak aerobic capacity (pVO 2 ) was 13.3 ± 3.3 mL.kg −1 .min −1 (51% of predicted value) in the MACE patients compared with 16.1 ± 3.7 mL.kg −1 .min −1 (67% of predicted value) in the patients with an absence of MACE. Forced vital capacity (FVC) was also decreased (71% vs. 82% predicted value) in the MACE patients, as was ventilatory reserve, with 35% of patients presenting with a ventilatory reserve under 25% (53% in "MACE" vs. 26% in "no MACE" patients; p = 0.06). Ventilatory inefficiency in the MACE patients was further ascertained by a high VEVCO 2 slope of 42.6 ± 6.6. A value of VEVCO 2 slope up to 35 was found in 95% of the MACE cases compared with 65% in the "no MACE" patients; p = 0.01.  [17,18]. † optimal cut-off points determined by Receiver Operating Characteristic (ROC) Curve analysis. Results are presented as mean ± standard deviation. Statistical significance was set at p < 0.05.
Figures 1A-C further illustrate the Kaplan-Meier curves of the primary endpoint MACE according to pVO2 (above and below 50% cut-off), FVC (above and below 70% cutoff) and the combination of pVO2 and FVC. Patients with reduced pVO2 (<50% predicted) presented with MACE more rapidly (mean survival free from MACE: 50 ± 13 months) compared with patients with pVO2 ≥ 50% (mean survival free from MACE: 90 ± 20 months) (log-rank p-value = 0.12) ( Figure 1A).   The same tendency was observed for the patients with reduced FVC (<70% predicted) with the corresponding mean MACE-free survival of 16 ± 2 months compared with the patients with FVC ≥ 70% (mean survival free from MACE: 99 ± 16 months) (log-rank p-value < 0.01) ( Figure 1B).
When pVO 2 and FVC were considered simultaneously, three patient profiles were outlined, presenting with significantly differing risk profiles (p < 0.01) when compared to a referent patient group (pVO 2 ≥ 50% and FVC ≥ 70%) with a mean survival of 167 ± 13 months ( Figure 1C).
The highest risk group consisted of patients presenting with both impaired pVO 2 (<50%) and FVC (<70%), with an HR for MACE of 25.60 (95% CI: 4.62-141.85, p < 0.01). Patients with the highest risk presented with a mean survival of 15 ± 6 months, and were thus subject to MACE onset, on average, 9-13 years earlier compared to patients with the lowest risk (referent group).
The quantification of the added predictive value of the functional parameters FVC (cut-off: 70%) and pVO 2 (cut-off: 50%) was in line with the above results. Indeed, the simultaneous consideration of FVC, pVO 2 and ATTR biomarker staging led to a significant average improvement of 35% in MACE risk prediction (Figure 3) compared to ATTR biomarker staging alone.
The resultant was the correct reclassification of 63% of the overall study population into more appropriate risk groups (p < 0.01), with 67% of the MACE patients reassigned a higher risk category (p < 0.01; Hosmer-Lemeshow goodness-of-fit test: p = 0.95) (Figure 4).

Discussion
With the advent of novel therapies for transthyretin cardiac amyloidosis, the identification of pertinent prognostic strategies with the most accurate ability to detect increased adverse event risk in patients is an important research pursuit. While clinical, echocardiographic and biological parameters are traditional severity markers in this chronic disease population, it has also been highlighted that impaired peak VO2 (pVO2) detains prognostic . Net Reclassification Improvement (NRI) for ATTR biomarker staging, peak VO 2 (% predicted) and FVC (% predicted). Model 1: ATTR biomarker; Model 2: ATTR biomarker + peak VO 2 (cut-off 50%) + FVC (cut-off 70%); Event defined as MACE; non-event defined as "no MACE". Abbreviations: MACE: composite of heart-failure-related hospitalization or all-cause death; ATTR: transthyretin amyloidosis; peak VO 2 : peak oxygen uptake; FVC: forced vital capacity; NRI: Net Reclassification Improvement. Optimal cut-off points for peak VO 2 and FVC determined by Receiver Operating Characteristic (ROC) curve analysis. Table Interpretation (

Discussion
With the advent of novel therapies for transthyretin cardiac amyloidosis, the identification of pertinent prognostic strategies with the most accurate ability to detect increased adverse event risk in patients is an important research pursuit. While clinical, echocardiographic and biological parameters are traditional severity markers in this chronic disease population, it has also been highlighted that impaired peak VO 2 (pVO 2 ) detains prognostic value, which might be optimized when considered with another functional parameter, lung volume restriction. We thus deemed that determining the potential ability of the combination of impaired pVO 2 and lung volume restriction to provide superior prognostic resolution would be a viable research endeavor, particularly compared with other established prognostic markers, such as ATTR biomarker staging.
Novel findings of our study are the following. Firstly, our study suggests that reduction in lung volume is a common feature in ATTR-CA patients (49%). Secondly, a restrictive spirometry pattern (FVC < 70% predicted) and reduced pVO 2 (<50% predicted) are identified as independent predictors of poor outcome (MACE), defined as either all-cause death or heart-failure-related hospitalization in ATTR-CA patients. MACE-free survival time significantly declines in patients with reduced FVC and impaired pVO 2 , with MACE onset occurring, on average, 9 to 13 years earlier compared to patients with a normal spirometry pattern and normal aerobic capacity. Thirdly, the combined use of spirometry (FVC) and pVO 2 improves risk stratification based on biomarker staging (NT-proBNP levels and estimated glomerular filtration rate) of ATTR-CA patients, with a 35% improvement in patient risk discrimination and 67% of the MACE patients reassigned into a higher risk category. IDI and NRI analyses further suggest that the greatest benefit in combining these three severity markers is moving patients with MACE into a higher risk group.
In accordance with earlier studies, we thus confirmed that reduced pVO 2 and increased VEVCO 2 slope seem to be strongly and independently predictive of MACE in ATTR-CA patients [5][6][7]9,10]. In addition, we also found that the presence of a restrictive spirometry pattern was associated with increased MACE risk in ATTR-CA, consistent with other authors reporting that the presence of a restrictive ventilatory pattern is predictive of all-cause and cardiovascular mortality [11][12][13]. Indeed, spirometry parameters can predict outcomes and improve risk stratification based on pVO 2 in chronic heart failure patients with reduced ejection fraction [13].
Pulmonary manifestations rarely dominate the clinical picture of ATTR cardiac amyloidosis [23,24]. In this amyloidosis type, lung involvement is most often a post mortem finding, as evidenced by diffuse amyloid deposit in alveolar septal and vessel walls [24,25]. However, some lines of evidence support the assertion that pulmonary involvement may be diagnosed ante mortem [23,26,27], with bone tracer scintigraphy and chest computed tomography commonly displaying abnormal pulmonary imaging [28][29][30]. In spite of the absence of clinically overt pulmonary involvement, a reduction in lung volume has been consistently reported using pulmonary functional parameters in ATTR-CA patients. In the present study, restrictive spirometry pattern was observed along with rapid and shallow breathing and ventilatory inefficiency during physical exertion. Moreover, it is important to note that the mechanisms of lung restriction often described in patients with heart failure [28,29] are not readily evident in ATTR-CA patients. No signs of lung edema, pleural effusion or increased cardiac size were observed in our patients. Nonetheless, our findings unequivocally suggest that behind the purported relative clinical silence, a restrictive spirometry pattern might be of certain clinical importance in ATTR-CA as corroborated by the observed significant association between MACE survival and lung volume restriction in our study population, as well as the high discriminating capacity of reduced forced vital capacity. Indeed, an optimal cutoff value for FVC of less than 70% of its predicted value defined a high to very high risk of MACE onset in our patients.

Study Limitations
An obvious limitation of the present investigation is its retrospective design, as well as the relatively small number of MACE events. Only 82 patients (31 events) were included, thus possibly resulting in a potential lack of statistical power to detect other potential factors with moderate effects on outcome. This small sample size is mainly explained by the fact that cardiac amyloidosis is a rare disease even in expert centers evaluating cardiopulmonary function in CA patients. Plethysmography was not available in all ATTR-CA patients. Plethysmography requires special features that are not available on a metabolic cart with VO 2 testing equipment. Hence, due to the lack of routine availability of plethysmography, advanced pulmonary function testing was not used to confirm our interpretations of airflow and ventilatory patterns resulting from basic spirometry testing. In contrast, spirometry remains one of the simplest and most widely available methods to assess pulmonary function.

Conclusions
To the best of our knowledge, this is the first exploratory study of the kind, assessing the combined prognostic value of pVO 2 , FVC and ATTR biomarker staging. The unequivocal changes in the three metric measures generally used to assess the predictive ability of a marker (AUC, IDI, NRI) all underline that combining biological (ATTR biomarker staging) and functional markers (pVO 2 and FVC) synergistically improved prognostic resolution in our cohort of patients with transthyretin cardiac amyloidosis. Our results suggest that the greatest benefit in combining these three severity markers is moving patients with adverse outcome (MACE) into a higher risk group, with the identification of subjects at greatest MACE risk with significantly greater accuracy compared with either marker alone. This additive predictive capacity of the three combined markers was consistent across our rather heterogeneous study population in terms of genetic variability (wild-type ATTR, different pathogenic ATTR variants) and disease severity (mild to severe disease states ascertained by the NYHA classification and ATTR biomarker staging). Our results need to be confirmed by future prospective investigations with a larger number of events, allowing for a more powerful multivariable analysis taking into account negative multisystem impact (i.e., cardiovascular, autonomic, pulmonary, renal, and skeletal muscle) of ATTR-CA. Such a multivariable scoring system would dramatically improve the ability to portend adverse event risk and allow more comprehensive risk prediction in these patients.
In light of the potential highly informative capacity of CPET and spirometry in predicting adverse outcome in ATTR-CA patients, these tests might prove to be highly useful tools, significantly contributing towards improved risk prediction, optimized patient monitoring and clinical decision making, as well as the timely introduction of newer-generation anti-amyloid therapies.

Acknowledgments:
The authors wholeheartedly thank the patients and medical teams of the four referral centers (France, Italy, USA) who contributed to this work. We convey special acknowledgement to the cardiology team of the University Hospital of Martinique at the origin of the present research initiative.

Conflicts of Interest:
The authors declare no conflict of interest.