DISCUSSION
The results of our study are in line with previous studies, and demonstrate that, following valve repair for degenerative disease, high postrepair gradients are of important clinical relevance. However, repair technique (in particular posterior MV leaflet resection) did not show a significant effect on postrepair gradients and no effect of postrepair gradients on the occurrence of late atrial fibrillation was observed.
The mechanism of elevated postrepair MV gradients is poorly understood, but is in our opinion related to fixation of the posterior annular perimeter length. Indeed, in an adult sheep model, Dagum et al. [9] have shown that cyclic changes in the MV annular area, reaching a maximum in late diastole, are correlated with length changes of the posterior annular perimeter. After implantation of a true-sized partial or complete flexible annuloplasty device, complete fixation of the MV annular area, posterior and anterior annular perimeters throughout the cardiac cycle was observed. Similarly, Bothe et al. [10] reported that MV annuloplasty results in a significant reduction of the maximal MV opening area, independent of the type of annuloplasty device used (partial or full ring; flexible, semi-rigid or rigid). Moreover, the Physio I ring (as well as the majority of other full rings implanted) did not impair posterior MV leaflet motion. Additionally, while the Physio I ring did increase the excursion of the annular and belly region of the anterior MV leaflet, it did not affect the excursion of the anterior leaflet at the leaflet edge. No effect on anterior leaflet motion was seen after the implantation of a partial Cosgrove-Edwards band (Edwards Lifesciences). Their results suggest that no diastolic flow obstruction due to changes in anterior leaflet motion is to be expected after implantation of the majority of annuloplasty devices tested.
In a study on 107 patients after MV repair for degenerative disease, Mesana et al. [5] suggested that full rings may be associated with higher postrepair gradients when compared to partial bands. Their results need to be interpreted with caution as the Duran ring (Medtronic Inc., Minneapolis, MN, USA), notoriously prone to significant pannus formation [11, 12], was used in most patients who underwent full ring implantation. As MV gradients were measured several years after the initial operation, the high gradients observed in this group are most likely due to pannus formation and not the type of annuloplasty device implanted. Murashita et al. [13] explored postrepair gradients in 1147 patients after MV repair for degenerative disease. In most patients, a standard 63 mm flexible band was implanted, irrespective of annular perimeter length or anterior leaflet size. In the first median follow-up period (124.5 days after operation), a mean gradient of 3.7 ± 1.6 mmHg was reported that declined to a mean gradient of 3.3 ± 1.8 mmHg in the second median follow-up period (600 days after operation). The gradients in the first median follow-up period seem somehow higher than early postrepair gradients observed in our population. On longitudinal analysis, a decline similar to the one observed by Murashita et al. was also observed in our population. While clearly limited, amongst others, by the unadjusted differences in patient populations, such observations challenge the alleged superiority of partial bands in terms of postrepair gradients. On the other side, excessive shortening of the posterior annular perimeter (resulting from downsizing described by the authors) is bound to have resulted in smaller MV areas than would have been achieved with a true-sized ring.
The results of our analyses suggest that the postrepair gradient is partially associated with patient characteristics (both patient age and body surface area likely reflect the relation of these parameters to systemic metabolic needs and hereto related cardiac output). Moreover, our results demonstrate that the choice of annuloplasty device used might affect postrepair gradients. This is likely correspondent to the differences in the ratio between the anterior-posterior diameter (reflecting anterior MV leaflet surface area to which the ring is sized) and the length of the ‘posterior perimeter’ of the annuloplasty ring between different annuloplasty devices. We are unaware of any studies demonstrating that the increased saddle shape of the Physio II ring is related to improved diastolic MV opening properties when compared to annuloplasty devices with less pronounced saddle-shape. Because of the lack of studies, we cannot exclude the possibility that the Physio II ring results in superior diastolic opening properties of the anterior leaflet that might explain our observation. For the time being, this should, however, be seen as a theory and properly designed studies will need to be conducted in the future.
In a recent meta-analysis, Mazine et al. [14] have speculated that in cases of posterior MV leaflet prolapse, chordal replacement techniques will result in lower MV gradients when compared to leaflet resection techniques. Our results failed to demonstrate any significant effect of posterior leaflet resection on postrepair gradient. This is likely related to the fact that we primarily used leaflet resection techniques to address excessive posterior MV leaflet tissue and structurally avoided excessive resection. Should this be avoided, no significant shortening of the posterior leaflet free edge, that could at least theoretically limit diastolic leaflet mobility, is to be expected. Functional MV stenosis should not present a reason not to perform adequate leaflet resection when this is indicated, an opinion previously emphasized by our group and others [15–18].
We did not observe any effect of elevated MV gradients on freedom from atrial fibrillation. On the contrary, elevated postrepair MV gradient was identified as a risk factor for late atrial fibrillation occurrence in recent studies by Kawamoto et al. [19] as well as Ma et al. [20]. We included only patients in sinus rhythm who did not undergo any ablation procedures, a characteristic that could provide an explanation for the differences observed between our results and those of Kawamoto et al. [19]. More importantly, we differentiated between atrial fibrillation and other types of atrial tachycardias as the mechanisms associated with the development of different atrial tachycardias after MV operations are known to fundamentally differ [7]. To adjust for potential unadjusted bias, future studies should include information on the type of incision made to expose the MV when exploring the effect of other factors on early and late atrial tachycardias. Previous studies also demonstrated that an elevated postrepair MV gradient impairs left atrial reverse remodelling [21]. Nevertheless, atrial fibrillation is most likely related to left atrial fibrosis that develops because of long-standing volume overload prior to the operation and is present even in patients in preoperative sinus rhythm [22]. More studies, with longer follow-up, are also needed to explore whether left atrial reverse remodelling in the presence of elevated postrepair gradient is impaired or only delayed. The results of our study do, however, suggest that the clinical burden of elevated postrepair gradient might be lower than previously suggested.
The observation that the risk of reintervention might be related to postrepair MV gradient has not previously been reported. The incidence of reoperation for MV stenosis after previous repair is known to be low [13, 23], but reflects only the patients most affected by this condition. Other studies have additionally shown that high postrepair gradients will result in decreased exercise tolerance and quality of life [3, 5]. We followed the established concepts of valve repair for degenerative disease that include annular remodelling and stabilization. As fixation of the maximal posterior perimeter length and hereto related maximal MV area are likely inevitable with any type of annuloplasty device used, elevated postrepair gradients might not be avoidable in all patients. Omitting annuloplasty device implantation is controversial, as this is known to result in a higher risk of recurrent MR [24]. Possibly, new annuloplasty device design and identification of patients in whom annular characteristics are sufficiently preserved to support valve sufficiency in the long term even without annular stabilization would help reduce the burden of this problem.

Limitations
This is a single-centre retrospective study with study limitations inherent to the study design. The results are applicable to the type of MV repair techniques described only and further studies are needed to evaluate the effect or other repair techniques (e.g. edge-to-edge repair) and annuloplasty devises on the occurrence of elevated postrepair gradients. During the study period, the MV repair techniques have evolved with, in particular, annular plication being performed less frequently. As the type of leaflet repair did not affect the sizing of the annuloplasty device implanted, no relevant effect on the results presented is to be expected. Moreover, the number of reinterventions was low and prevented us from exploring the risk factors for MV reintervention specified to the indication for reintervention (i.p. recurrent MR versus elevated MV gradient). While we failed to identify a relation between postrepair MV gradient and survival or atrial fibrillation, we cannot exclude the possibility that a correlation does exist, but we were unable to detect it due to an insufficient number of patients and events. Our results should thus be seen as hypothesis-generating and will need to be confirmed by future studies. We have performed cubic spline analyses, but failed to identify a cut-off value that would ease the identification of patient a risk for complications related to elevated postrepair MV gradient in the clinical setting. Identification of a cut-off value should be pursued in future studies. Nevertheless, in line with previous studies, our results do support the efforts aimed at securing the lowest possible postrepair MV gradient.