Natural history and clinical outcomes of de novo functional tricuspid regurgitation after the Ross procedure

Abstract

Objective

To characterize the natural history, predictors, and clinical outcomes of de novo severe functional tricuspid regurgitation (TR) following the Ross procedure.

Methods

At a single institution, 699 consecutive adults without preoperative severe TR undergoing the Ross procedure between 2011 and 2025 were reviewed. De novo severe TR was defined as TR grade ≥3 within 6 months postoperatively. Propensity score overlap weighting was used to balance 21 covariates between patients who developed de novo severe TR and those who did not. Weighted generalized estimating equations were used to analyze longitudinal TR trajectories. Predictors were identified using LASSO regression. Survival and reintervention were compared using weighted Kaplan-Meier and Aalen-Johansen methods.

Results

De novo severe TR occurred in 61 patients (8.7%), appearing at a median of 5 days postoperatively. The groups demonstrated significantly different TR trajectories (P for interaction <.001), with rapid spontaneous improvement in the de novo severe TR group. TR severity decreased to less than severe in 49 of the 61 patients (80.3%) in 2 years. Independent predictors of de novo severe TR included female sex (odds ratio [OR], 4.79; 95% confidence interval [CI], 1.99-11.56), diabetes (OR, 2.53; 95% CI, 1.09-5.88), and preoperative TR grade per unit increment (OR, 2.30; 95% CI, 1.26-4.20). At 10 years, freedom from death was 100% in the TR group versus 93.4% in the control group (P = .26), and cumulative reintervention on any valve was 3.0% versus 5.3% (P = .62).

Conclusions

De novo severe TR after the Ross procedure is common but transient, resolving spontaneously without any adverse impact on survival or need for reintervention. These findings support a conservative management strategy with echocardiographic surveillance.

Graphical Abstract

Tricuspid regurgitation (TR) trajectory showing resolution of de novo severe TR over time.

Central Message.

De novo functional tricuspid regurgitation after the Ross procedure occurred in 8.7% of patients but was transient and did not affect long-term outcomes.

Perspective.

This study demonstrates that de novo severe tricuspid regurgitation after the Ross procedure is observed in 9% of patients, appears early (median 5 days), and resolves spontaneously in more than 90% of patients at 3 years. Conservative management is appropriate given comparable long-term survival and freedom from reintervention.

The Ross procedure, involving autologous pulmonary valve transfer to the aortic position with pulmonary homograft reconstruction, provides excellent long-term outcomes in appropriately selected patients.1, 2, 3 Contemporary series demonstrate outstanding survival approaching that of the age-matched general population, with low rates of structural valve deterioration and freedom from anticoagulation.^4,5

As cumulative experience with the Ross procedure grows, attention has extended beyond autograft and homograft durability to the broader spectrum of postoperative sequelae. Functional tricuspid regurgitation (TR) is a well-recognized complication of left-sided cardiac surgery and carries prognostic significance.6, 7, 8 After mitral valve surgery, progressive TR is associated with heart failure, hepatic congestion, and reduced survival, often prompting reintervention.^9,10 Recently published clinical guidelines now emphasize earlier TR surgery.¹¹

Whether these observations extend to aortic valve surgery, and specifically to the Ross procedure, remains unknown. The Ross procedure differs fundamentally from conventional aortic valve replacement (AVR), involving a more extensive operation with typically longer cardiopulmonary bypass (CPB) time, autograft harvesting that disrupts the right ventricular (RV) outflow tract, and implantation of a pulmonary homograft. Each of these elements could predispose to RV dysfunction and secondary TR through mechanisms distinct from those operating after isolated valve replacement.

The appropriate management of postoperative TR, whether conservative observation or early intervention, depends on understanding its trajectory and prognostic impact. The present analysis characterizes the incidence, temporal evolution, predictors, and long-term clinical consequences of de novo severe functional TR in a large contemporary Ross cohort.

Methods

Study Design and Population

This retrospective cohort study used prospectively collected data from the Montreal Heart Institute cohort. A total of 712 consecutive adult patients (age ≥18 years) who underwent the Ross procedure between January 2011 and January 2025 were identified. After excluding patients with preoperative severe TR (grade ≥3; n = 1) and those developing severe TR after 6 months (n = 12), 699 patients were included in the analysis. This study was approved by the Montreal Heart Institute’s Research Ethics Board (IRB 2026-3636; approved February 26, 2026), with a waiver of individual patient consent owing to the study’s retrospective nature. Patient selection followed STROBE guidelines (Figure 1).¹²

Figure 1.

Flow diagram showing participant selection and follow-up according to the STROBE guidelines. Of 712 patients who underwent the Ross procedure between 2011 and 2025, 699 met the inclusion criteria and were analyzed. Patients were classified by the development of severe tricuspid regurgitation (TR) (≥3) within 6 months postoperatively.

Definitions, Follow-up, and Outcomes

Data were collected from the institution's electronic health records and from a prospectively maintained surgical database. The primary outcome was de novo severe functional TR, defined as TR grade ≥3 (moderate-severe or greater) occurring within 6 months of surgery in patients with preoperative TR grade <3. One patient was excluded because of his preoperative moderate-severe TR. Patients developing severe TR beyond 6 months (n = 12) were excluded from the analysis, as this late onset represents a distinct pathophysiology from perioperative de novo TR. Individual chart review confirmed that all 12 of these patients had granular echocardiographic assessment within 12 months without severe TR.

Resolution of TR was defined as improvement from severe TR (grade ≥3) to TR grade <3 on echocardiographic follow-up. This includes improvement to none (grade 0), trivial (grade 0.5), mild (grade 1), or moderate (grade 2) TR. TR severity was graded on a scale of 0 to 4 according to contemporary guidelines, integrating qualitative and semiquantitative parameters.¹³

Clinical follow-up was 90% complete (follow-up index with 2-year landmark), totaling 3626 patient-years. The median follow-up was 4.5 years (interquartile range [IQR], 1.8-8.3; maximum, 14.5 years). For the outcomes analysis, patients were stratified by the occurrence of de novo severe TR. Secondary outcomes included all-cause mortality and reintervention on any valve.

Statistical Analysis

The analyses addressed 2 distinct questions requiring different methodological approaches.

Characterizing patients with de novo TR

Unweighted analyses were used to describe patients developing de novo TR and identify predictors. Baseline and operative characteristics were compared using the Mann-Whitney U test for continuous variables and the Fisher exact test for categorical variables. Early postoperative outcomes (≤30 days) are presented descriptively, as these events precede or occur concurrently with TR detection and reflect the perioperative course rather than consequences of TR. Independent predictors were identified using least absolute shrinkage and selection operator (LASSO) logistic regression with 10-fold cross-validation. Variable stability was assessed via 500 bootstrap resamples, retaining predictors selected in ≥50% of iterations. Final odds ratios (ORs) were derived from multivariable logistic regression with variables rescaled to clinically interpretable units. Model calibration was assessed using the Hosmer-Lemeshow goodness-of-fit test and graphical calibration plots.

Estimating the effect of de novo TR on long-term outcomes

To assess whether de novo TR affects late outcomes, valid comparison requires adjustment for confounding. Patients developing TR differ systematically from those who do not (Table 1). Propensity score overlap weighting was used, which balances measured confounders while retaining all patients and focusing inference on patients with clinical equipoise between groups.¹⁴ The propensity model incorporated 21 baseline covariates. Balance was assessed using standardized mean difference (with a value <0.1 considered adequate) (Figure E1).

Table 1.

Baseline characteristics

Variable	De novo TR (N = 61)	Control (N = 638)	P value	SMD∗	De novo TR (OW)	Control (OW)	SMD†
Age, y, mean ± SD	48.4 ± 11.3	48.8 ± 11.9	.66	0.035	48.8 ± 11.2	48.6 ± 10.7	0.013
Female sex, n (%)	39 (63.9)	174 (27.3)	<.001	0.788	(61.2)	(51.1)	0.203
BMI, kg/m, mean ± SD†	28.1 ± 6.2	28.4 ± 5.2	.62	0.051	28.2 ± 6.2	28.1 ± 5.9	0.017
BSA, m2, mean ± SD	1.8 ± 0.2	1.9 ± 0.2	<.001	0.497	1.8 ± 0.2	1.9 ± 0.2	0.087
Hypertension, n (%)	23 (37.7)	241 (37.8)	1.00	0.001	(38.7)	(36.9)	0.036
Dyslipidemia, n (%)	23 (37.7)	206 (32.3)	.47	0.113	(37.9)	(32.2)	0.119
Diabetes mellitus, n (%)	10 (16.4)	55 (8.6)	.077	0.235	(14.1)	(17.6)	0.097
COPD, n (%)	4 (6.6)	34 (5.3)	.57‡	0.051	(6.8)	(6.8)	0.001
CKD, n (%)	3 (4.9)	18 (2.8)	.42‡	0.108	(5.5)	(3.2)	0.111
Coronary artery disease, n (%)	4 (6.6)	86 (13.5)	.16‡	0.231	(8.0)	(6.3)	0.064
PVD, n (%)	0 (0.0)	5 (0.8)	1.00‡	0.126	(0.0)	(0.4)	0.084
Current/former smoker, n (%)	14 (23.0)	127 (19.9)	.71	0.071	(24.6)	(22.1)	0.060
Atrial fibrillation, n (%)	2 (3.3)	18 (2.8)	.69‡	0.026	(3.4)	(4.1)	0.033
Previous cardiac surgery, n (%)	15 (24.6)	82 (12.9)	.019	0.303	(21.0)	(21.5)	0.013
NYHA class III–IV, n (%)	18 (29.5)	102 (16.0)	.013	0.325	(24.9)	(31.3)	0.143
Endocarditis, n (%)	3 (4.9)	35 (5.5)	1.00‡	0.025	(4.3)	(6.0)	0.075
STS mortality score, %, mean ± SD	1.1 ± 0.8	0.9 ± 0.6	<.001	0.337	1.0 ± 0.7	1.1 ± 0.7	0.046
EuroSCORE II, %, mean ± SD	1.1 ± 0.5	1.0 ± 0.4	.090	0.209	1.1 ± 0.4	1.2 ± 0.8	0.167
Etiology, n (%)
AS	53 (86.9)	535 (83.9)	.66	0.085	(87.1)	(85.4)	0.050
AR	8 (13.1)	145 (22.7)	.12	0.252	(13.0)	(21.9)	0.235
Mixed AS/AR	13 (21.3)	165 (25.9)	.53	0.107	(22.0)	(27.1)	0.117
Morphology, n (%)
Bicuspid valve	36 (59.0)	384 (60.2)	.97	0.024	(60.0)	(54.3)	0.116
Unicuspid valve	7 (11.5)	155 (24.3)	.035	0.338	(13.1)	(21.1)	0.214
Tricuspid aortic valve	9 (14.8)	62 (9.7)	.31	0.153	(13.7)	(12.1)	0.045
Quadricuspid valve	0 (0.0)	4 (0.6)	1.00‡	0.112	(0.0)	(0.5)	0.101
Prosthesis	9 (14.8)	33 (5.2)	.006	0.322	(13.2)	(11.9)	0.039
AV gradient, mm Hg, mean ± SD	42.5 ± 17.9	43.1 ± 28.4	.91	0.027	43.0 ± 17.9	43.6 ± 22.7	0.027
LVEDD, mm, mean ± SD	47.7 ± 9.7	49.7 ± 8.6	.032	0.214	47.6 ± 9.1	48.5 ± 8.3	0.114
LVESD, mm, mean ± SD	32.7 ± 8.3	32.2 ± 8.6	.69	0.062	32.5 ± 8.0	31.1 ± 8.9	0.164
IVS, mm, mean ± SD	11.1 ± 3.0	11.8 ± 2.9	.058	0.211	11.0 ± 2.7	11.4 ± 2.7	0.129
LVEF, %, mean ± SD	58.6 ± 9.3	60.0 ± 7.9	.49	0.160	58.9 ± 9.0	58.3 ± 9.7	0.063
Aortic annulus, mm, mean ± SD	23.2 ± 3.1	24.7 ± 3.2	<.001	0.502	23.4 ± 3.1	23.4 ± 3.0	0.024
Aortic sinus, mm, mean ± SD	31.6 ± 4.1	33.6 ± 5.5	.008	0.408	31.8 ± 4.0	32.4 ± 5.5	0.121
Sinotubular junction, mm, mean ± SD	26.8 ± 3.7	29.1 ± 4.9	.003	0.523	26.8 ± 3.6	28.3 ± 4.9	0.337
Mitral regurgitation ≥2+, n (%)	5 (8.2)	14 (2.2)	.019	0.271	(6.5)	(4.2)	0.103
TR ≥2+, n (%)	6 (9.8)	11 (1.7)	<.001	0.350	(5.4)	(10.9)	0.199
Preoperative TR grade, median [IQR]	0.5 [0.5-1]	0.5 [0-0.5]	<.001	0.505	0.5 [0-0.5]	0.5 [0-0.5]	0.108
Donor age, y, mean ± SD	42.4 ± 9.2	43.3 ± 9.1	.49	0.091	42.4 ± 9.3	43.4 ± 8.7	0.103

Postweighting values represent overlap-weighted percentages. TR, Tricuspid regurgitation; SMD, standardized mean difference; OW, overlap weighting; BMI, body mass index; BSA, body surface area; COPD, chronic obstructive pulmonary disease; CKD, chronic kidney disease; PVD, peripheral vascular disease; NYHA, New York Heart Association; STS, Society of Thoracic Surgeons; AS, aortic stenosis; AR, aortic regurgitation; AV, aortic valve; LVEDD, left ventricular end-diastolic diameter; LVESD, left ventricular end-systolic diameter; IVS, interventricular septum; LVEF, left ventricular ejection fraction; IQR, interquartile range.

^∗SMD before OW.

^†SMD after OW.

^‡Fisher exact test.

Figure E1.

Standardized mean differences (SMDs) for baseline covariates before (black dots) and after (red dots) overlap weighting (OW). The vertical dashed line represents the balance threshold of 0.1 absolute SMD, indicating adequate covariate balance.

Notably, patient selection for the Ross procedure requires meticulous preoperative assessment of right heart function, as the operation depends on a competent pulmonary valve and preserved right ventricle, both of which contribute to the challenges posed by the prolonged CPB and cross-clamp times. Patients with high pulmonary pressures, regardless of a normal right ventricle, or patients with RV dysfunction are not candidates for this operation, inherently narrowing the range of preoperative RV dysfunction within this cohort. Although quantitative RV parameters (eg, tricuspid annular plane systolic excursion [TAPSE], RV systolic velocity [RV S′]) were not systematically reported on preoperative echocardiography, the propensity model incorporated New York Heart Association (NYHA) functional class, left ventricular ejection fraction, preoperative TR grade, and pulmonary hypertension as clinical surrogates for disease severity and right-sided hemodynamic burden.

Longitudinal TR trajectories were compared using weighted generalized estimating equation (GEE) models with an exchangeable correlation structure, testing the group-by-time interaction. Freedom from death was estimated using weighted Kaplan-Meier analysis and compared with expected survival obtained from Statistics Canada 2023 life tables (Ederer II method).¹⁵ The cumulative incidence of reintervention was estimated using a weighted Aalen-Johansen method with death as a competing risk. All analyses were performed using Python 3.12, with a 2-sided P < .05 for significance. NYHA functional class was compared between the 2 groups at each follow-up time point using ordinal logistic regression. A weighted GEE model assessed whether NYHA trajectory differed between the groups over time.

As a sensitivity analysis, the entire analytic pipeline was repeated using TR ≥2 (moderate or greater) as the threshold for de novo severe TR rather than TR ≥3 (moderate-severe or greater). This involved re-derivation of cohort groups, propensity score estimation with overlap weighting, GEE trajectory modeling, survival analysis, and predictor identification. A sensitivity analysis also was conducted for the rate of resolution, in which resolution was defined as less than moderate TR.

Results

Study Population

Among the 699 patients, 61 (8.7%) developed de novo severe TR within 6 months following surgery. Baseline characteristics are presented in Table 1. Patients in the de novo TR group were more frequently female (63.9% vs 27.3%; P < .001), had a smaller body surface area (1.8 m² vs 1.9 m²; P < .001), more frequent prior cardiac surgery (24.6% vs 12.9%; P = .02), higher NYHA functional class (P = .013), and higher preoperative TR grade (median, 0.5 [IQR, 0.5-1.0] vs 0.5 [IQR, 0.0-0.5]; P < .001). Although this TR grade difference is clinically difficult to interpret, preoperative TR grade was maintained in the LASSO predictor model, suggesting that even subclinical differences in baseline TR may identify patients with greater susceptibility to perioperative tricuspid dysfunction. Beyond the 61 patients with early de novo severe TR (≤6 months), 12 additional patients developed severe TR later during follow-up (median onset, 6 years; range, 12 months to 12 years). Individual review confirmed echocardiographic assessment without severe TR within 6 months postoperatively in all 12 cases. The late onset and distinct temporal profile suggest a mechanism unrelated to perioperative insult. Including these patients, the overall incidence of severe TR at any point after the Ross procedure was 10.4% (n = 73/699).

Echocardiographic follow-up surrogates for right heart function were available in the full cohort. Left ventricular ejection fraction converged by 1 year (59.0 ± 6.4% vs 59.6 ± 5.1%; P = .48) and peak pulmonary gradient was consistently lower in the de novo TR group throughout follow-up, arguing against chronic RV outflow obstruction. In a subset with preoperative cardiac magnetic resonance imaging (MRI) (n = 177; 25.3%), RV ejection fraction and volumes were comparable in the 2 groups, and postoperative MRI confirmed preserved RV systolic function (Table E1) .

Operative characteristics are summarized in Table 2. CPB time (mean, 192.7 ± 53.1 minutes vs 181.0 ± 41.0 minutes; P = .23) and cross-clamp time (mean, 168.8 ± 40.2 vs 161.1 ± 32.0 minutes; P = .27) were similar in both groups. Patients with de novo TR received aortic annuloplasty rings less frequently (11.5% vs 25.1%; P = .03).

Table 2.

Operative characteristics

Variable	De novo TR (N = 61)	Control (N = 638)	Weighted P value
Autograft size, mm, mean ± SD	23.9 ± 2.4	24.7 ± 2.2	.001
Ring annuloplasty, n (%)	7 (11.5)	160 (25.1)	.026
Ascending aorta replacement, n (%)	26 (42.6)	291 (45.6)	.98
Hemiarch replacement, n (%)	3 (4.9)	42 (6.6)	1.00∗
Pulmonary homograft size, mm, mean ± SD	28.4 ± 1.2	28.7 ± 1.3	.063
Concomitant procedures, n (%)	33 (54.1)	363 (56.9)	.77
CABG, n (%)	4 (6.6)	33 (5.2)	.53∗
MV repair, n (%)	3 (4.9)	14 (2.2)	.16∗
TV repair, n (%)	0 (0.0)	0 (0.0)	—
CPB time, min, mean ± SD	192.7 ± 53.1	181.0 ± 41.0	.23
Cross-clamp time, min, mean ± SD	168.8 ± 40.2	161.1 ± 32.0	.27

TR, Tricuspid regurgitation; CABG, coronary artery bypass grafting; MV, mitral valve; TV, tricuspid valve; CPB, cardiopulmonary bypass.

^∗Fisher exact test.

Timing and Natural History of de Novo TR

Perioperative echocardiographic assessment consistently demonstrated normal right-sided valve function. Among patients with de novo severe TR, detection occurred at a median of 5 days postoperatively (mean, 12.8 days; range, 1-176 days). The median detection at 5 days reflects the timing of the discharge echocardiogram. All detected de novo severe TR instances were strictly functional. Figure E2 details cohort- wide TR trajectories as well as individual de novo severe TR trajectories. Weighted GEE analysis demonstrated significantly different TR trajectories between the groups (P for interaction <.001). The de novo TR group demonstrated rapid spontaneous improvement over time (interaction coefficient, −0.060; 95% confidence interval [CI], -0.086 to −0.033), converging toward the control group trajectory during follow-up (Figure 2, A).

Figure E2.

Detailed tricuspid regurgitation (TR) trajectory. A, TR grade distribution over time of the entire cohort. B, Individual trajectories of de novo moderate-severe TR patients.

Figure 2.

A, Longitudinal trajectory of tricuspid regurgitation (TR) grade by group. Model-predicted mean TR grade trajectories from a generalized estimating equations model with overlap weighting. The model demonstrates significant improvement in TR grade over time in the de novo severe TR group (red) while the control group (blue) remained stable, with a statistically significant group-by-time interaction (P < .001). B, Cumulative incidence of resolution of TR calculated with the Aalen-Johansen method. Shaded areas represent 95% confidence intervals.

The cumulative incidence of resolution (TR <3) was 49.1% (95% CI, 36.6%-63.1%) at 1 year (Figure 2, B). TR severity decreased to less than severe (grade <3) in 80.4% of the patients (95% CI, 68.7%-89.9%) with de novo severe TR after 2 years, and in 57 of 61 patients (93.4%; 95% CI, 85.4%-98.5%) after 3 years. The remaining patients had persistent moderate-severe TR at their last follow-up, although none underwent tricuspid intervention. Baseline characteristics were similar in the resolved and persistent subgroups. No patient in either subgroup died, required tricuspid intervention during follow-up, or required high-dose diuretics. Functional class was comparable (median, NYHA class I) in both groups at all postoperative time points; however, hospitalization for heart failure was required in 2 of 12 persistent patients (17%) compared to 0 of 49 resolved patients. One of these patients was admitted in the third postoperative week for sudden left-sided heart failure with volume overload that improved with diuretics. The second patient was admitted 1 month after an operation for right-sided heart failure that resolved following intravenous diuretic infusion. Although never reaching statistical significance, likely owing to a small sample size, persistent patients had a lower pulmonary peak gradient at all postoperative time points.

Bootstrap-validated LASSO regression identified 4 independent predictors of de novo severe TR (Figure 3). The multivariable model demonstrated good discrimination (C-statistic = 0.792; 95% CI, 0.728-0.851) and excellent calibration (Hosmer-Lemeshow χ² = 4.40; P = .819), indicating that predicted probabilities closely matched observed event rates across all risk strata (Figure E3). Female sex conferred the highest risk (OR, 4.79; 95% CI, 1.99-11.56; P < .001), followed by diabetes mellitus (OR, 2.53; 95% CI, 1.09-5.88; P = .03) and preoperative TR grade per unit increment (OR, 2.30; 95% CI, 1.26-4.20; P = .007). Higher ejection fraction was protective, with each 10% increment reducing the odds by 34% (OR, 0.66; 95% CI, 0.46-0.94; P = .02).

Figure 3.

Forest plot showing predictors retained after LASSO logistic regression with 10-fold cross-validation and stability assessment across 500 bootstrap iterations. Odds ratios reflect the final multivariable model, with all variables scaled to clinically interpretable units.

Figure E3.

Model performance. A, Calibration plot showing observed versus expected event proportions across deciles of predicted risk (χ² = 4.40; P = .819). B, Receiver operating characteristic curve demonstrating model discrimination (C-statistic = 0.792; 95% confidence interval, 0.728-0.851).

Clinical Outcomes

Early outcomes are presented in Table 3. Operative mortality was similar in both groups (0% vs 0.2%; P = 1.00). Patients with de novo TR demonstrated higher rates of blood transfusion (55.7% vs 37.1%; P = .005) and acute kidney injury necessitating renal replacement therapy (14.8% vs 5.5%; P = .010). Hospital length of stay was longer in the TR group (mean, 8.3 ± 9.7 days vs 5.9 ± 4.7 days; P < .001), as was ICU length of stay (mean, 3.7 ± 3.6 days vs 2.2 ± 2.8 days; P < .001).

Table 3.

Postoperative characteristics

Variable	De novo TR (N = 61)	Control (N = 638)	De novo TR (OW)	Control (OW)	P value
30-day/in-hospital mortality, n (%)	0 (0.0)	1 (0.2)	(0.0)	(0.0)	1.00∗
Stroke, n (%)	0 (0.0)	2 (0.3)	(0.0)	(0.2)	1.00∗
Reoperation for bleeding, n (%)	1 (1.6)	11 (1.7)	(0.8)	(2.1)	1.00∗
AKI necessitating RRT, n (%)	9 (14.8)	35 (5.5)	(12.4)	(7.7)	.010
Atrial fibrillation/flutter, n (%)	19 (31.1)	147 (23.0)	(29.6)	(26.4)	.21
New permanent pacemaker, n (%)	1 (1.6)	6 (0.9)	(2.0)	(0.6)	.47∗
Sternal wound infection, n (%)	0 (0.0)	2 (0.3)	(0.0)	(0.3)	1.00∗
Prolonged ventilation >24 h, n (%)	3 (4.9)	18 (2.8)	(5.5)	(4.0)	.42∗
Reintubation, n (%)	5 (8.2)	12 (1.9)	(8.2)	(2.9)	.009
Blood transfusion, n (%)	34 (55.7)	237 (37.1)	(54.3)	(47.5)	.005
Length of stay, d, mean ± SD	8.3 ± 9.7	5.9 ± 4.7	8.5 ± 10.1	6.8 ± 5.7	<.001
ICU stay, d, mean ± SD	3.7 ± 3.6	2.2 ± 2.8	3.7 ± 3.7	2.7 ± 3.6	<.001
Predischarge echocardiographic parameters
iEOA, cm2/m2, mean ± SD	1.0 ± 0.4	1.2 ± 0.6	1.0 ± 0.4	1.3 ± 0.6	.12
Mean aortic gradient, mm Hg	4.8 ± 2.6	5.6 ± 2.7	4.6 ± 2.4	5.6 ± 2.7	.015
Peak aortic gradient, mm Hg, mean ± SD	8.7 ± 4.5	9.4 ± 4.5	8.4 ± 4.2	9.3 ± 4.5	.16
AR ≥2+, n (%)	0 (0.0)	2 (0.3)	(0.0)	(0.3)	1.00∗
Peak pulmonary gradient, mm Hg, mean ± SD	5.1 ± 3.6	6.5 ± 3.2	5.1 ± 3.6	5.9 ± 2.9	<.001
Aortic sinus, mm, mean ± SD	31.9 ± 3.6	32.7 ± 4.1	31.7 ± 3.5	31.8 ± 4.3	.32
LVEF, %, mean ± SD	55.3 ± 10.5	56.9 ± 7.4	55.2 ± 10.4	56.4 ± 8.3	.67

TR, Tricuspid regurgitation; OW, overlap weighting; AKI, acute kidney injury; RRT, renal replacement therapy; ICU, intensive care unit; iEOA, indexed effective orifice area; AR, aortic regurgitation; LVEF, left ventricular ejection fraction.

^∗Fisher exact test.

Freedom from late death was excellent in both groups (Figure 4, A). At 10 years, survival was 100% in the de novo TR group, compared with 93.4% (95% CI, 90.2%-95.6%) in controls (P = .26). Both groups demonstrated survival comparable to population survival (95.3% at 10 years).

Figure 4.

A, Kaplan-Meier curves showing freedom from late death following the Ross procedure in patients with de novo severe tricuspid regurgitation (TR; red), controls (blue), and the general population (black dashed line). B, Cumulative incidence of reintervention stratified by TR status, estimated using the Aalen-Johansen method. Shaded areas represent 95% confidence intervals.

The cumulative incidence of reintervention was similarly low in both groups (Figure 4, B). At 10 years, the rate of reintervention was 3.0% (95% CI, 0.3%-12.4%) in the de novo TR group versus 5.3% (95% CI, 3.0%-8.7%) in controls (P = .62). Freedom from reintervention was 97.6% overall. Notably, no patient in either group required isolated or combined tricuspid intervention.

NYHA functional class improved substantially in both groups following surgery (Figure E4). Despite higher baseline NYHA class in the de novo TR group (P = .013), postoperative functional status was similar (P = .433 at 0-1 year). GEE analysis demonstrated no significant difference in NYHA trajectory between the groups (P for interaction = .261), with both groups achieving and maintaining excellent functional recovery.

Figure E4.

Distribution of New York Heart Association (NYHA) functional class over time. Stacked bar charts showing the proportion of patients in each NYHA class before surgery and at different follow-up intervals for the control and de novo severe tricuspid regurgitation groups.

Sensitivity Analysis

When the threshold was lowered to TR ≥2, the de novo group expanded from 61 to 154 patients, with 510 controls (vs 638 in the primary analysis). The GEE group × time interaction remained significant (P < .001; primary analysis, P < .001), confirming the robustness of the trajectory findings. Survival (log-rank P = .220) and reintervention (P = .185) results were consistent with the primary analysis (Table E2, Figure E5). Predictors of de novo TR also were consistent throughout both analyses. A sensitivity analysis also was performed for time to resolution by redefining resolution as TR <2 (moderate). At 2 years, the resolution rate dropped from 80.4% (95% CI, 68.7%-89.9%) to 55.8% (95% CI, 42.7%-69.9%) (Figure E5).

Figure E5.

Sensitivity analysis of tricuspid regurgitation (TR) trajectory. A, TR trajectory with de novo TR defined as >2. B, TR trajectory with de novo TR defined as >1. C, Resolution of de novo TR, with resolution defined as <3. D, Resolution of de novo TR, with resolution defined as <2.

Discussion

This analysis characterizes de novo severe functional TR following the Ross procedure. The principal findings are (1) de novo severe TR occurred in 8.7% of patients and manifested early (at a median of 5 days postoperatively); (2) TR resolved spontaneously in 80.3% of affected patients at 2 years; (3) female sex, diabetes, preoperative TR severity, and reduced ejection fraction were independent predictors; and (4) after adjustment for confounding, de novo TR had no impact on long-term survival, freedom from reintervention, or functional class.

The 8.7% incidence of de novo severe TR in this cohort is comparable to the 6% to 20% incidence reported after conventional AVR, suggesting that the underlying pathophysiology is shared across cardiac surgical procedures rather than specific to the Ross operation. Prolonged CPB, ischemia-reperfusion injury, and perioperative volume shifts likely represent the predominant mechanisms across all forms of aortic valve surgery.16, 17, 18 However, the Ross procedure introduces additional theoretical risk factors, including disruption of the RV outflow tract during autograft harvesting and homograft implantation. These particularities warrant caution in determining whether postoperative TR follows a similarly benign course in this population. The slight difference likely reflects the younger, healthier population selected for the Ross procedure, with fewer comorbidities and better-preserved ventricular function. The stability of TR in the control group mirrors findings after conventional AVR, suggesting the Ross procedure does not inherently provoke progressive TR in most patients.¹⁹ The unique physiology of the Ross procedure, which restores a living aortic root with more physiologic hemodynamics than a prosthesis, also may confer protective effects on RV adaptation, though this hypothesis requires further investigation.

All instances of de novo severe TR in our cohort were confirmed to be functional in mechanism, based on echocardiographic findings of central regurgitant jets and annular dilation. In contrast, an iatrogenic structural injury would manifest as an eccentric jet (particularly directed posteriorly toward the homograft suture line), along with restricted leaflet motion of the septal or posterior leaflet, visible tethering or abnormal chordal geometry, and failure to improve on serial imaging. In fact, the strongest evidence against iatrogenic structural injury is the spontaneous resolution of TR in most affected patients. The early appearance (predominantly during the index hospitalization) and subsequent improvement strongly suggest an acute, reversible RV maladaptation rather than progressive pathology. Several mechanisms likely contribute, including CPB-induced myocardial stunning, acute shifts in interventricular septal geometry following aortic root replacement, and transiently altered RV loading conditions.^20,21 Unlike the progressive annular dilation driven by chronic pulmonary hypertension or atrial fibrillation–related remodeling, TR following the Ross procedure appears to be a functional consequence of temporary ventricular insult or volume overload that resolves with myocardial recovery.²² The higher rate of acute kidney injury necessitating renal replacement therapy seen in patients with de novo TR (14.8% vs 5.5%; P = .010) likely reflects the hemodynamic consequences of acute RV dysfunction, including systemic venous congestion and reduced renal perfusion pressure.²³ Importantly, this perioperative morbidity did not translate into long-term adverse outcomes, further supporting the transient nature of the underlying RV insult.

The predictors identified here offer insight into patient vulnerability. Female sex emerged as the strongest predictor, with 18.3% of all females developing de novo severe TR, compared with 4.5% of males. Sex-stratified analysis demonstrated that this vulnerability was not attributable to more advanced disease at presentation; females had similar functional class, significantly better-preserved ejection fraction, and fewer comorbidities. But anatomic differences remained, specifically smaller cardiac dimensions, including smaller aortic annular diameter, body surface area, and left ventricular cavity size. Table E3 details sex-based differences. This is consistent with growing evidence that sex-based differences in tricuspid disease are driven in part by annular geometry and sex-specific remodeling responses.^24,25 The association with diabetes may involve subclinical cardiomyopathy. Diabetic cardiomyopathy may impair RV reserve during perioperative stress²⁶ and has been independently linked to RV dysfunction in heart failure populations.²⁷ The protective effect of higher ejection fraction aligns with the hypothesis that de novo TR represents transient RV dysfunction; patients with better baseline function may be more resilient to surgical insult. The predictive value of preoperative TR grade suggests that patients with higher baseline TR grade, even though less than moderate, are more susceptible to developing severe TR under perioperative stress. Consistent with this observation, Kang and colleagues²¹ have associated preoperative mild TR with the development of significant TR after isolated AVR.

Patients with de novo severe TR were managed conservatively with medical optimization. This included intravenous diuretic therapy during hospitalization for volume control and RV preload/afterload optimization, followed by a standardized 10-day course of oral diuretics after discharge. Myocardial protection evolved during the study period. Intermittent cold blood cardioplegia was used from 2011 to 2016, after which del Nido cardioplegia was adopted. In patients with aortic stenosis, arrest was initiated via antegrade delivery through a dual-lumen catheter at 12 mL/kg, supplemented by direct ostial delivery (300 mL) after 30 minutes, with retrograde maintenance via a coronary sinus catheter (300 mL every 30 minutes). These strategies showed no difference in the rate of TR presentation. In patients with aortic regurgitation, arrest was initiated via retrograde delivery, with direct ostial cannulation if arrest was not achieved, followed by the same retrograde maintenance protocol. Systemic hypothermia was maintained at 34 °C throughout CPB. In cases of circulatory arrest for extended associated aortic replacement, deeper hypothermia (25-28 °C) was used.

Importantly, causal analysis using propensity score weighting demonstrated that de novo TR did not adversely affect long-term outcomes. After balancing confounders, there was no difference in 10-year survival (100% vs 93.4%; P = .26) or reintervention (3.0% vs 5.3%; P = .62). These findings support a conservative management strategy of watchful waiting, contrasting with functional TR after mitral valve surgery, in which progression is common and earlier intervention may be warranted.^11,22,28 In the Ross population, early postoperative TR appears to be transient and unlikely to warrant tricuspid intervention.

Limitations

This study has limitations inherent to its retrospective, single-center design. Patient selection for Ross procedure and postoperative management are institution-specific, which may limit the generalizability of our findings. The reliance on clinically driven echocardiographic follow-up rather than on protocol-mandated imaging may introduce surveillance bias, although this likely would be bias toward detecting more TR in symptomatic patients. Despite rigorous propensity score weighting, the possibility of unmeasured confounding cannot be excluded. Quantitative RV functional parameters (TAPSE, RV S′, tricuspid annular diameter) were not incorporated into the propensity score model. The evaluation of RV global longitudinal strain and fractional area change were not systematically captured in patient reports at our institution. However, preoperative cardiac MRI in a subset of patients demonstrated comparable RV ejection fraction and RV volume between groups, and the inherent selection criteria for the Ross procedure exclude patients with RV dysfunction, narrowing the potential for unmeasured RV confounding. Of note, perioperative RV parameters were not routinely reported on intraoperative transesophageal echocardiography when RV function was qualitatively normal. Future prospective studies should incorporate quantitative RV assessment, including tricuspid annular dimensions, TAPSE, and RV S′, on preoperative, and perioperative, and surveillance echocardiography. The absence of a core echocardiographic laboratory for standardized TR grading is a further limitation; grading variability across individual readers may have introduced measurement imprecision. Moreover, although the median time to TR detection was 5 days, this value largely reflects the timing of the discharge echocardiogram rather than the true moment of TR onset. Despite rigorous follow-up, the exact onset of TR remains obscured by the limited temporal resolution inherent to echocardiographic surveillance. Finally, while follow-up was robust (90% with a 2-year landmark; median, 4.5 years), the low event rate in the de novo TR group limits the statistical power for detecting small differences in rare outcomes. A multicenter prospective registry would provide greater certainty regarding long-term prognosis.

Conclusions

De novo severe TR following the Ross procedure occurs in 8.7% of patients, manifests early in the postoperative period, and resolves spontaneously in the majority. Female sex, diabetes, preoperative TR severity, and reduced ejection fraction are independent predictors. After adjustment for confounding, de novo TR does not adversely impact long-term survival or freedom from reintervention. Conservative management with echocardiographic surveillance is the appropriate strategy for this transient phenomenon.

Webcast

You can watch a Webcast of this AATS meeting presentation by going to: https://www.aats.org/resources/natural-history-and-clinical-o-12216.

Conflict of Interest Statement

The authors reported no conflicts of interest.

The Journal policy requires editors and reviewers to disclose conflicts of interest and to decline handling or reviewing manuscripts for which they may have a conflict of interest. The editors and reviewers of this article have no conflicts of interest.

Appendix E1

Table E1.

RV surrogate parameter analysis

Variable	De novo TR, mean ± SD	Control, mean ± SD	P value
Preoperative RV imaging
PAP, mm Hg	49.4 ± 15.1	45.9 ± 13.8	.759
PR grade	0.3 ± 0.3	0.3 ± 0.3	.723
TR grade	0.7 ± 0.5	0.5 ± 0.4
Pulmonary peak gradient, mm Hg	4.6 ± 5.7	4.8 ± 4.7	.054
Pulmonary HTN (binary)	0.1 ± 0.3	0.0 ± 0.1	.004
LVEF, %	58.6 ± 9.3	60.0 ± 7.9	.488
LVEDD, mm	47.7 ± 9.7	49.7 ± 8.6	.032
LVESD, mm	32.7 ± 8.3	32.2 ± 8.6	.691
LV mass, g	113.2 ± 50.7	116.6 ± 44.7	.223
RVEF (MRI), %∗	56.5 ± 6.6	57.4 ± 8.4	.560
RVEDV (MRI), mL∗	129.3 ± 46.5	139.2 ± 44.4	.328
RVESV (MRI), mL∗	59.9 ± 26.2	61.3 ± 25.9	.748
Postoperative RV imaging
RVEF, %∗	55.1 ± 4.5	55.4 ± 6.2	.846
RVEDV, mL∗	128.8 ± 27.8	152.6 ± 40.8	.037
RVESV, mL∗	58.1 ± 13.9	68.7 ± 22.9	.112
PV regurgitation fraction, %∗	5.2 ± 5.9	6.7 ± 8.6	.591

RV, Right ventricular; PAP, pulmonary artery pressure; PR, pulmonary regurgitation; TR, tricuspid regurgitation; HTN, hypertension; LVEF, left ventricular ejection fraction; LVEDD, left ventricular end-diastolic diameter; LVESD, end-systolic diameter; LV, left ventricular; RVEF, right ventricular ejection fraction; MRI, magnetic resonance imaging; RVEDV, right ventricular end-diastolic volume; RVESV, right ventricular end-systolic colume; PV, pulmonary valve.

^∗MRI data are incomplete. Preoperative and postoperative de novo TR group completeness was n = 15 and n = 12, respectively, and control group completeness was n = 159 and n = 129, respectively.

Table E2.

Sensitivity analysis

Parameter	TR ≥3 (primary)	TR ≥2 (sensitivity)
Cohort
Total cohort, n	699	664
De novo group, n	61	154
Control group, n	638	510
Excluded (preoperative TR ≥ threshold), n	1	18
Excluded (late developers), n	12	30
Covariate balance (OW)
Covariates with SMD <0.1 (after), n/N	18/21	18/21
Maximum SMD after weighting	0.235	0.138
GEE trajectory
Group effect (coefficient)	0.945	0.759
Group × time interaction (coefficient)	−0.06	−0.059
P value for interaction	<.001	<.001
Survival
Deaths, de novo/control, n/N	0/13	1/11
Log-rank P value for mortality	.260	.220
Log-rank P value for reintervention	.615	.185
Significant predictors (in order of importance)	Female sex, diabetes, preoperative TR grade (per grade), ejection fraction	Female sex, preoperative TR grade (per grade)

TR, Tricuspid regurgitation; OW, overlap weighting; SMD, standard mean deviation; GEE, generalized estimating equation.

Table E3.

Sex-based analysis of the de novo severe TR group

Variable	Females (N = 39)	Males (N = 22)	P value
Demographics, median [IQR]
Age, y	51.0 [38.0-57.0]	52.5 [46.2-56.8]	.448
BMI, kg/m2	27.0 [23.0-32.0]	26.5 [25.2-30.0]	.391
BSA, m2	1.7 [1.6-1.8]	2.0 [1.8-2.1]	<.001
Comorbidities, n (%)
Hypertension	13 (33.3)	10 (45.5)	.507
Diabetes	6 (15.4)	4 (18.2)	1.000
Smoking	4 (10.3)	10 (45.5)	.003
CAD	0 (0.0)	4 (18.2)	.014
Atrial fibrillation	0 (0.0)	2 (9.1)	.126
Previous cardiac surgery	11 (28.2)	4 (18.2)	.539
Functional status and risk, median [IQR]
NYHA class	2.0 [2.0-3.0]	2.0 [2.0-3.0]	.762
STS mortality, %	1.0 [0.7-1.3]	0.7 [0.5-1.1]	.031
EuroSCORE II, %	1.1 [0.7-1.4]	0.9 [0.6-1.2]	.070
Valve pathology, n (%)
Aortic stenosis	35 (89.7)	18 (81.8)	.443
Aortic regurgitation	5 (12.8)	3 (13.6)	1.000
Endocarditis	1 (2.6)	2 (9.1)	.293
Preoperative EEG parameters, median [IQR]
LVEF, %	60.0 [60.0-65.0]	60.0 [47.2-62.5]	.034
Mean AV gradient, mmHg	38.7 [31.5-52.5]	42.4 [36.1-52.0]	.532
AVA, cm2	0.9 [0.6-0.9]	0.9 [0.8-1.1]	.258
Preoperative TR grade	0.5 [0.5-1.0]	0.5 [0.5-1.0]	.786
Preoperative PR grade	0.5 [0.0-0.5]	0.0 [0.0-0.5]	.151
MR grade	0.5 [0.0-1.0]	0.5 [0.5-1.0]	.547
PAP, mmHg	49.0 [40.0-58.0]	50.0 [44.0-55.5]	1.000
LVEDD, mm	43.5 [41.0-47.5]	50.0 [44.0-56.0]	.030
LVESD, mm	31.0 [25.0-35.0]	32.0 [30.0-41.5]	.228
IVS, mm	10.5 [9.0-12.2]	11.0 [10.0-12.0]	.364
Aortic annulus, mm	22.5 [21.0-23.2]	24.2 [23.0-26.0]	.011
Sinus of Valsalva, mm	30.5 [28.0-32.8]	32.5 [31.0-33.2]	.034
LV mass, g	85.0 [76.0-126.0]	115.0 [90.0-162.0]	.056
Operative data
CPB time, min, median [IQR]	176.0 [151.0-220.0]	178.5 [148.5-234.8]	.910
Cross-clamp time, min, median [IQR]	161.0 [136.5-191.5]	152.0 [137.0-201.5]	.898
Autograft size, mm, median [IQR]	23.0 [22.0-24.0]	25.0 [24.0-26.0]	<.001
Homograft size, mm, median [IQR]	28.0 [28.0-29.0]	29.0 [28.0-30.0]	<.001
Annuloplasty ring, n (%)	3 (7.7)	4 (18.2)	.240
TV repair, n (%)	1 (4.3)	0 (0.0)	1.000
Donor age, y, median [IQR]	42.0 [34.2-49.0]	45.0 [42.0-50.0]	.090

TR, Tricuspid regurgitation; IQR, interquartile range; BMI, body mass index; BSA, body surface area; CAD, coronary artery disease; NYHA, New York Heart Association; STS, Society of Thoracic Surgeons; EEG, electrocardiogram; LVEF, left ventricular ejection fraction; AV, aortic valve; AVA, aortic valve area; MR, mitral regurgitation; PAP, pulmonary artery pressure; LVEDD, left ventricular end-diastolic diameter; LVESD, left ventricular end-systolic diameter; IVS, interventricular septum; LV, left ventricular; CPB, cardiopulmonary bypass; TV, tricuspid valve.