All patients underwent three-dimensional CT scans before surgery to measure the Com–Com length, PAL, and the minimum distance between the mitral annulus and LCX. A 320-row CT detector (Aquilion ONE/GENESIS Edition, Canon Medical Systems, Otawara, Japan) was used for all scans before surgery of the patients. The CT scan parameters were as follows: 320 rows × 0.5 mm; rotation time, 0.275 s; volume scan; tube voltage, 120 kV; tube current, 800–1,000 mA. Three-dimensional coronary arteries were reconstructed at mid-diastole (75% of the R–R cycle) with a slice thickness of 0.5 mm, a slice interval of 0.25 mm, and a soft reconstruction kernel (FIRST Cardiac Mild). Synapse Vincent software (Synapse Vincent, Fujifilm Medical, Tokyo, Japan) was used to analyze CT data. The coronary arteries were automatically drawn using this software. The mitral annulus was drawn as a three-dimensional line by manually plotting the hinge point of the mitral leaflet. The distance from the mitral annulus to the closest point on the LCX artery wall was measured and expressed as the minimum coronary annular distance (mCAD) (Figure 1A). All image creation and measurements were performed by cardiac surgeons (NK and KNi), after which a radiologist (SI) reassessed the accuracy of the measurements.

All patients underwent transthoracic echocardiography (TTE) in our echocardiography laboratory the day before surgery. The MR grade was evaluated by measuring the color Doppler jet area using the Doppler-derived volumetric method or the proximal isovelocity surface area method. For statistical analysis in the present study, the grading of MR or TR was scored as follows: none = 0, none to mild = 0.5, mild = 1, mild to moderate = 1.5, moderate = 2, moderate to severe = 2.5, and severe = 3 (7). The mitral valve area, calculated by pressure half time, was calculated using the following formula: 220 /pressure half time.

All patients underwent MV repair using the robot. Our loop technique has been described in a previous study (8, 9). The loop set consisted of a felt pledget and CV4 expanded polytetrafluoroethylene (ePTFE) suture loops (Gore-Tex®, WL Gore & Associates, Flagstaff, AZ, USA). The number of loops (usually two or three) depended on the extent of the prolapsed area. The ePTFE suture needles were passed through the corresponding papillary muscle, with the loop set anchored 3–5 mm below the tip of the muscle. The sutures were then passed through the second felt pledget and secured with several knots. In addition, the loop was affixed to the edge of the prolapsing segment with a 5-0 Prolene® (Ethicon, Cincinnati, OH, USA) figure-of-8 suture.

We used the loop technique for MV repair and a semi-rigid ring (Carpentier-Edwards Physio Ring II, Edwards Lifesciences, Irvine, CA, USA) for annuloplasty in all 100 patients. All annuloplasty procedures were performed using the continuous wrapping suture technique (Figure 1B). Ring size was determined based on the Com–Com distance or anterior leaflet area using a sizer (Figure 2A). We placed an interrupted suture at the bilateral fibrous trigons using a 90-cm CV4 PTFE suture (Gore-Tex, Flagstaff, AZ, USA). The threads were sutured continuously at the mitral annulus from the bilateral trigone, effectively wrapping the ring in place. The ring was inserted into a continuous suture without passing through it. In cases where the mCAD was shorter than 3 mm, continuous sutures were performed with a fine pitch and shallow bite.

The plication ratio of the posterior mitral annulus was expressed as the ratio of the posterior side of the prosthetic ring (ring PAL) to the native PAL. The accuracy of ring size selection was expressed as the ratio of the anterior side of the prosthetic ring (ring Com–Com) to the native Com–Com length. The ring PAL, ring Com–Com (Figure 2B), and size Com–Com (Figure 2A) for each ring size are shown in Figure 2C.

Postoperatively, we followed up with the patients through hospital visits, outpatient clinic appointments, and telephone surveys. All patients completed follow-up, which ranged from 90 to 1,608 days (median, 336 days). Postoperative echocardiographic follow-ups were performed at our hospital 7–14 days after surgery and every 3–6 months after discharge. The follow-up duration ranged from 30 to 1,429 days, with a median of 457 days.

Descriptive statistics for categorical variables were reported as absolute values and percentages, while continuous variables were presented as means and standard deviations. The association between the PAL plication rate and echocardiographic data was analyzed using the Spearman rank correlation coefficient. Receiver operating characteristic (ROC) curves and areas under the curve were assessed using standard methods. The optimal cutoff value was defined as that with the highest sum of sensitivity and specificity. Event-free rates were estimated using the Kaplan–Meier product-limit method. Statistical analysis was performed using R version 3.3.2 (R Foundation for Statistical Computing, Vienna, Austria). Statistical significance was set at P < 0.05.

The patient profiles are presented in Table 1. Most patients had hypertension. Almost all patients were MR with degenerative (82%) or Barlow's-like disease (14%). Preoperative TTE showed slightly dilated diastolic dimensions and left atrial volume index (LAVI), while the left ventricular ejection fraction remained within the normal range.

The distribution of PAL across all patients is presented in Table 2. The PAL of most patients ranged from 91 to 130 mm. Patients with Barlow’s disease had longer PAL (120 ± 14 mm) and Com–Com (34 ± 2.8 mm) measurements than degenerative patients (p = 0.00114 and p = 0.00342, respectively). There were no significant differences in PAL and Com–Com among the prolapse lesion types. In addition, 48% of patients had an mCAD of less than 3 mm and 6% of patients had an mCAD of less than 1 mm. The mCAD was the shortest in the left-dominant type among the three types of coronary systems.

The perioperative and postoperative echocardiographic data are summarized in Table 3. One-fourth of the patients underwent concomitant procedures, and six patients (6%) experienced systolic anterior motion. Of the six patients with systolic anterior motion, two required additional height reduction under the second pump run, resulting in the recovery of systolic anterior movement. Other four patients recovered from systolic anterior movement through volume loading. The plication rate of the PAL was 0.68 ± 0.069 and did not differ depending on each ring size. Based on preoperative Com–Com measurements, the accuracy of the selected ring size was 0.99 ± 0.08. There were neither perioperative deaths nor cases of low cardiac output syndrome. One patient required re-exploratory thoracotomy due to bleeding. The MR scores at both postoperative periods and the most recent periods were 0.49 ± 0.47 and 0.89 ± 0.67, respectively. No patient required reoperation. Two patients had moderate or severe MR, and there were no instances of recorded ring detachment or LCX obstruction.

ROCs were used to evaluate the optimal cutoff value of plication rate × ring size for determining a postoperative mean transmitral mean pressure gradient of 5 mmHg or more (Figure 3A). A PAL × ring size of 19.4 provided the maximum sum of sensitivity and specificity for predicting the postoperative mean transmitral pressure gradient (81% and 86%, respectively). The area under the curve was 0.83. The plication rate × ring size corrected by body surface area (BSA) showed a cutoff value of 11.1, with the maximum sensitivity and specificity for predicting the postoperative mean transmitral pressure gradient of 82% and 86%, respectively. The area under the curve was 0.82.

Mean transmitral pressure gradient for each ring size and each plication rate are shown in Table 4.

For smaller ring sizes (26 and 28 mm), the mean transmitral pressure gradient was high when the posterior plication rate was less than 0.65. On the other hand, for ring sizes larger than 34 mm, the mean transmitral pressure gradient remained low even when the suture ratio was less than 0.6.

As for the (ring PAL/PAL) × ring size, the mean transmitral pressure gradient was higher (3.5 ± 1.8 mm Hg) when the (ring PAL/PAL) × ring size was less than 19.0.

Figure 3B shows the Kaplan–Meier analysis, which demonstrated that freedom from moderate or severe MR was not different between the two groups above and below the posterior plication rate × ring size of 19.3 (p = 0.73), with event-free rates of 97% vs. 96% at 1 year and 97% vs. 96% at 3 years, respectively. Differences in the PAL plication rate did not affect MR recurrence in the mid-term periods.

Lower PAL plication resulted in a higher recovery rate of EF during the most recent periods (p = 0.0268) (Figure 4).

This study has several limitations. First, our study had a retrospective design, which included only a small number of patients. Nevertheless, our data were obtained from a single reference surgical team committed to attempting valve repair in almost all cases of degenerative prolapse. Furthermore, all surgeries were performed by the first author (YT) using a robot with an excellent operative view. Second, measurement bias in preoperative CT measurements and intraoperative measurements may occur. Third, we performed postoperative enhanced three-dimensional CT only in patients whose postoperative creatine kinase-MB (CK-MB) levels were >100 IU. Fortunately, no patients with coronary obstruction underwent postoperative enhanced three-dimensional CT. Fourth, it would be important to compare the wrapping suture technique with a semi-rigid ring not only to other types of technique including interrupted suturing technique but also to other types of rings. Fifth, if the MV orifice area, calculated by the continuity equation, was determined using the stroke volume measured in the LV outflow tract divided by the integral of the trans mitral valve velocity during diastole, the relationship between the posterior plication ratio and the effective valve orifice area may be more accurately calculated (20). As this was a retrospective study, only pressure half time was available, but it would be desirable to measure MV area in future studies.

Finally, we did not compare the plication rate using semi-rigid rings with that using other ring types. Although the mitral repair technique employed was the loop technique without leaflet resection in all degenerative cases, the number of loops and additional techniques were not consistent.