The Impact of Age in Prosthesis-patient Mismatch on Long-term Survival after Aortic Valve Replacement: in-vitro versus in-vivo Values

Aim: We studied the effect of age on survival in the setting of prosthesis-patient mismatch (PPM). Study Design: Retrospective single surgeon practice. Place and Duration of Study: Cardiothoracic department, Mater Dei Hospital Malta, between January 1995 and December 2014. Methodology: 572 consecutive patients undergoing aortic valve replacement (AVR) were divided into four age groups and followed up for a maximum of 20 years (mean 8.2). Date of death was derived from the National Statistics Office. PPM was classified according to defined criteria, and calculated according to manufacturers’ tables (in-vitro) and from in-vivo values published by independent researchers. The impact of age and PPM on long-term survival was studied using the Cox proportional hazard model. Original Research Article Manché et al.; JAMPS, 9(4): 1-8, 2016; Article no.JAMPS.28381 2 Results: Mean in-vitro derived indexed effective orifice area (EOAi) was significantly higher than invivo EOAi (1.04±0.22 vs 0.93±0.16, p=0.000) and incidence of PPM was lower using in-vitro criteria (moderate 18.0% vs 24.1%, p=0.01, severe 1.9% vs 4.7%, p=0.008). For patients with mismatch the odds of dying (in-vitro vs in-vivo) was increased by 9.2% vs 38.1%, with moderate PPM 7.6% vs 30.9%, and with severe PPM 85.7% vs 69.7%. The odds of dying increased with age (by 7-8% for every year) and PPM severity. Age was a significant predictor of survival but PPM was not. For every 0.1unit increase in EOAi the risk of dying decreased by 8.0% (in-vitro) and 8.7% (in-vivo). Conclusion: Age is a significant predictor of survival times, with the odds of dying increasing by about 7% for every additional year. Long-term survival hazard was increased by PPM but the effect was not significant. When EOAi is analysed as a continuous variable it significantly effects survival.


INTRODUCTION
Rahimtoola introduced the concept of prosthesispatient mismatch (PPM), denoting an effective orifice area of the prosthetic valve too small for the patient's body size, in 1978 [1].Valve manufacturers provide the surgeon with effective orifice area (EOA) tables based on in-vitro testing of prostheses [2,3].However in-vivo testing by independent researchers has underlined the prevalent overestimation of EOA in these tables [4,5] and recommendations for EOA estimation have since been revised accordingly [6].
In turn manufacturers have developed valves designed for supra-annular implantation where the valve size label, based on the tissue annular diameter (TAD) is equal to the internal orifice diameter (IOD), in contrast to valves designed for intra-annular implantation where the IOD is smaller than the TAD [7].Other design features such as a reduced sewing ring and external mounting of pericardial tissue are geared towards providing a larger EOA [8].
The mitigation of PPM is based on the premise that sub-optimal haemodynamics result in adverse clinical outcomes.Studies have demonstrated persistent left ventricular hypertrophy and dysfunction with consequent poorer functional class and quality of life [9], increased incidence of late cardiac events [10], and reduced durability of bioprosthetic valves [11].The combined deleterious effects of these factors on long-term survival is, however, still controversial [12,13].
Advancing age impacts negatively on immediate, medium-and long-term survival after aortic valve replacement (AVR), but the direct effect of PPM remains unresolved [14][15][16].In this paper we studied the effect of age and PPM on long-term survival applying both the manufacturers' in vitro EOA's as well as in-vivo values derived from independent researchers.

METHODOLOGY
572 consecutive patients in a single-surgeon practice (61% male, mean age 65.1±11.4)requiring surgical AVR /+coronary artery bypass grafting (CABG) between January 1995 and Dec 2015 were enrolled in the study.Patients requiring other additional surgery or transcatheter aortic valve implantation were excluded.Data was collected prospectively and date of death was derived from the National Statistics Office.The population was divided into four age groups of comparable size: age15-59 (n=148), 60-67 (n=145), 68-74 (n=149), >74 (n=130).The maximum follow-up period was 20 years, with a mean of 8.2 and a median of 7.6 years.

All
patients underwent surgery under normothermic cardiopulmonary bypass and myocardial protection was with antegrade cold cardioplegia.CABG was performed in the standard fashion and the graft of first choice was the internal thoracic artery.
The cut-off point for xenografts implantation was set at age 70 and this was adhered to in 93% of patients below 70 and 95% of patients above 70.The choice of valves evolved with time in line with newer models with the promise of superior haemodynamics (Table 1).No root-enlargement procedures were undertaken.
Moderate PPM was defined as an indexed effective orifice area (EOAi, effective orifice area per m 2 body surface area) of 0.65-0.85cm 2 /m 2 and severe as <0.65 cm 2 /m 2 , and was calculated according to tables (in-vitro) supplied by the valve manufacturers and also from in-vivo values published by independent researchers (Table 2).

Statistical Analysis
The impact of age and vitro/vivo PPM classification on long-term survival was studied using the Cox proportional hazard model.This model combines a baseline hazard function of time with an exponentiated term including a linear combination of the predictors (age and vitro/vivo PPM classification).Parameters are estimated by maximizing the partial likelihood function with respect to the parameters and the hazard ratios are the exponential values of these parameter estimates.

RESULTS
185 patients died and the remaining 387 were right censored.Fig. 1  Table 3 shows that age is a significant predictor of survival times and that for every incremental year the odds of dying increase by around 7%.Moreover, for patients with mismatch the odds of dying (in-vitro and in-vivo) were respectively 9.2% and 38.1% higher compared to patients with no PPM, but the increase was not statistically significant.
For patients with moderate and severe PPM, the odds of dying were respectively 7.6% and 85.7% higher compared to patients with no PPM, using in-vitro values, and 30.9% and 69.7% higher using in-vivo values.These increases were not statistically significant (Table 4).
Age was significant predictor of survival times, however PPM was not significant.This is attributed mainly to the low incidence of mismatch, particularly for severe PPM.When we consider EOAi as a continuous parameter rather than classifying it into three categories, namely: no PPM, moderate PPM and severe PPM, we find that a higher EOAi was associated with a significantly decreased hazard ratio of dying, both when using the in-vitro data and well as with the in-vivo data (Table 6).
For every 0.1-unit increase in EOAi the hazard of dying, rather than surviving, decreases by 8.0% (in-vitro criteria) and by 8.7% (in-vivo criteria) given that other effects are kept fixed.Age remains a significant predictor of long-term survival in that for every year increase in the patient's age the hazard ratio of dying rather than surviving increases by 2% to 3%.
The change in use of valve models from the beginning of 2002 resulted in a significant decrease in PPM by in-vitro criteria ( ).
From our analysis we conclude that age was a significant predictor of survival, whereas PPM failed to reach a statistically significant effect on long-term survival.This situation applied for moderate and severe mismatch, for all valve sizes used, and for calculations based on the manufacturers' EOA's as well as those provided by independent researchers.In contrast, when analyzing EOAi as a continuous variable we find that it exerts a significant incremental effect on long-term survival.(1) = 6.027, p = 0.014

DISCUSSION
In a large meta-analysis of the impact of PPM on long-term survival, Head et al concluded that allcause mortality was significantly increased and cardiac-related mortality non-significantly increased [27].The authors stressed the importance of PPM prevention especially in younger patients who often receive a mechanical prosthesis, causing a higher negative impact on survival.A number of studies included in this analysis failed to demonstrate a significant effect of PPM.Only one of the 34 studies had a longer mean follow-up of 9.1 years vs 8.2 in our study [28] and a second was comparable, with a median of 7.3 years vs 7.6 in our study [29].Both studies showed no effect (Frapier 2000 HR 0.49 [95% CI 0.25, 0.96], Tsutsumi 2008 HR 0.88 [95% CI 0.34, 2.29]), suggesting that the duration of follow-up may be an important factor.
Our study demonstrated PPM to be more prevalent in younger patients and we therefore analyzed the interaction of age and PPM on long-term survival.While age was an independent significant predictor of curtailed survival, PPM was not.Particularly with older patients, the duration of follow-up is largely determined by the age at operation.The interaction of age and follow-up duration may play an important role in determining long-term outcomes, a longer follow-up favoring age as the significant predictor.Although PPM is known to result in persistent left ventricular hypertrophy and accelerated xenograft dysfunction, the combined effect of these factors with advancing age renders the effect on long-term survival more complex.Our study suggests that age, and indirectly, follow-up duration, are more important than PPM in determining long-term survival.
When we analyze EOA as a continuous variable, rather than categorizing it into no PPM, moderate PPM and severe PPM, we find that it significantly affects long-term survival.A contributing factor to this finding is the low incidence of PPM, particularly of the severe category, limiting its statistical significance.Surgeons should be cognizant of this phenomenon when implanting an aortic prosthesis.
Although the manufacturers' declared EOA's resulted in a lower incidence of mismatch when compared to calculations using independent researchers' values, both incidences and degrees of mismatch failed to significantly affect long-term survival.In spite of criticisms in the literature leveled at the manufacturers' tables, our study showed them to be clinically valid [5].We did not include our own EOAi's measured after valve implantation, only calculating this value from published data, as this represents the situation encountered during valve replacement.We recommend that both the manufacturers' as well as independent researchers' data is available in the operating theatre to help the surgeon in deciding on valve model and size.
In line with current recommendations important consideration should be given to the prevention of PPM whenever possible, by implanting valves of an adequate size and with superior haemodynamic performance.When this is not possible root enlargement may be contemplated, although this procedure increases operative complexity and has not been shown to benefit long-term survival [30].We effected a change to newer generation prostheses in 2002 with a resultant decrease in the incidence of PPM.However other studies have shown that this strategy does not affect mortality [27].

CONCLUSION
Age is a significant predictor of long-term survival, with the odds of dying increasing by about 7% for every additional year.EOAi exerts a significant incremental effect on survival with every 0.1 unit increase decreasing the risk of dying by 8%.Long-term survival hazard was increased by PPM but the effect was not significant, irrespective of the severity of the PPM, the valve size implanted, and the source of the EOA values, whether provided by the valve manufacturers or independent researchers.

Fig. 1 .
Fig. 1.Kaplan-Meier survival curves for the four age groups

Table 1 . Valves implanted during the study period 1995-2002* 2002-2015 **
Size refers to valve size implanted.Source (in-vivo) is derived from independent researchers' data in postoperative patients.Source (in-vitro) is derived from manufacturers' data from laboratory tests.Reference column refers to publications from which the EOA values are derived shows the Kaplan-Meier survival curves for the four age groups.

Table 3 . Cox regression relating survival time to age and PPM (in-vitro vs in-vivo) Predictor Parameter estimate SE p value Hazard Ratio 95% lower CI higher
SE: standard error, CI: confidence interval

Table 6 . Cox regression relating survival time to age and EOAi: in-vitro and in-vivo calculation
df: degrees of freedom associated with each parameter estimate.Wald test: used to test the true value of the parameter, based on the sample estimate