Comparison of Short- and Medium-Term Clinical Outcomes between Transradial Approach and Transfemoral Approach in a High-Volume PCI Heart Center in China

Background Transradial approach (TRA) outweighed transfemoral approach (TFA) in acute coronary syndrome patients because the former has better short-term outcomes in high-volume percutaneous coronary intervention (PCI) centers. Our study was one of the limited studies specifically in comparing the short- and medium-term effects of TRA and those of TFA in patients undergoing elective PCIs. Methods A total of 21,242 patients who underwent elective PCI with stent implantation were included. Using propensity score methodology, 1,634 patient pairs were matched. Major clinical outcomes and PCI-related complications between TRA and TFA were compared. Results In the propensity score-matched patients, the rates of in-hospital net adverse clinical events, which included death, myocardial infarction (MI), target vessel revascularization (TVR), stroke, and major bleeding, were much lower with TRA than with TFA (1.8% vs. 3.9%, P < 0.001). This difference was mainly due to the lower rate of major bleeding (0.6% vs. 1.8%, P < 0.001) and the decreased rate of MI (1.1% vs. 1.9%, P = 0.060). PCI-related dissection and thrombosis were similar between the TRA and TFA groups (both P > 0.05). Meanwhile, one-year incidence rates of major adverse cardiovascular events, which included death, MI, and TVR, were also similar (4.1% vs. 4.9%, P = 0.272) in TRA and TFA. Multivariable regression analyses showed that TRA was an independent predictor of the low rate of in-hospital net adverse clinical events (odds ratio, 0.53; 95% confidence interval, 0.40 to 0.71), but not of major adverse cardiovascular events at one-year follow-up (hazard ratio, 1.01; 95% confidence interval, 0.96 to 1.06). Conclusions In patients undergoing elective PCI, TRA patients had lower rates of in-hospital net adverse clinical outcomes compared with TFA patients. TRA might be recommended as a routine approach in high-volume PCI hospitals for elective PCIs.


Introduction
It has been 20 years since the first percutaneous coronary intervention (PCI) through transradial approach (TRA) was successfully performed by Dr. Kiemeneij and Laarman. [1] Since then, this method has been increasingly adopted because of its superior features (lower vascular complications, shorter hospitalization, and better experience for patients) compared with the transfemoral approach (TFA). [2][3][4][5] The application rate of TRA varies across countries. In Asia and some European countries, TRA is a routine approach. By contrast, the application rate of TRA is very low in the US. [6,7] China is among the first countries to use the TRA strategy and to maintain its practice. A cross-national survey in China showed that TRA accounted for 56.3% of all routes for PCI in 2007, and this percentage increased to 76.1% in 2011. [8] In Fuwai hospital, TRA has been used since 2000. More than 10 000 PCIs are performed each year, and TRA has accounted for approximately 90% of all routes. Learning curves exist for TRA beginners, however, most of the interventionists are systematically trained and have accumulated a great amount of experience in implementing TRA. [9] The RIVAL trial (trial in RadIal Vs. femorAL access for coronary intervention) suggested that institution with high volume of PCI had better 30 d outcomes with TRA compared with TFA in acute coronary syndrome patients. [10] While very few studies have tested that in patients undergoing elective PCI. Thus, a comparison of major procedural and clinical outcomes between TRA and TFA during hospitalization and at one-year follow-up was performed. Meanwhile, a propensity score methodology was used to decrease the disparities.

Patient selection and data collection
This study was conducted in a single institution-Fuwai hospital, the largest heart center in China. The annual amount of PCI in Fuwai hospital was more than 10,000 after year 2011, and the proportion that uses TRA in PCI is increasing. The statistical method of the study was based on a post-hoc analysis of a prospective database. From 1 June 2006 to 30 April 2011, a total of 23,389 patients who have undergone PCI with stent implantation were included in the study. The flow chart for patient selection was shown in Fig 1. Finally, 21,242 patients were included in the analysis. The angiographic data was downloaded from the digital database of the catheter laboratory. The baseline data and the in-hospital outcomes were extracted from the medical charts. A group of systematically trained medical students worked on the data extraction. Four senior fellows constituted the quality control committee, in which one person was in charge of adjudicating ambiguous endpoints and the others were responsible for data inspection. Informed consents were obtained upon patient admission, and follow-up checkups were performed at 6 months and 1 year after discharge. The protocol of this study was approved by the Ethics Committee of Fuwai hospital, and patient information was anonymized and deidentified prior to analysis.

The procedure
TRA is preferred in our institution. Under certain conditions however, doctors are still willing to choose the femoral route. Such conditions included a failed Allen's test, a weak or non-palpable pulse, and a history of coronary artery bypass surgery. Doctors had good expertise both with TRA and TFA. 36 interventionists performed 19,363 procedures (with available name of the operator in the medical record) during approximately 5 years. Among them, when 13 operators who performed less than 15 procedures were excluded, then each operator performed about 150 procedures per year. Using that as benchmark, 14 operators had procedures less The Outcomes of Transradial Approach PCI than 150 annually; 4 operators had procedures between 150 to 300 annually; and 5 operators had procedures more than 300 annually.
At the start of the procedure, 1 ml 1% lidocaine was subcutaneously injected for local anesthesia, and the puncturing needle and the sheath were used to build the route. The hemostasis method was easier for TRA than for TFA. In TRA patients, hemostasis was achieved by manual compression of the puncturing site, followed by clamp placement over the artery. In TFA patients, sandbag compression was demanded after manual compression, and patients were required to stay in bed for hours before ambulation.

Endpoint definition and follow-up checkups
The primary endpoint was defined as in-hospital net adverse clinical events (NACE), which included all causes of death, myocardial infarction (MI), target vessel revascularization (TVR), stroke, and major bleeding. The secondary endpoint was defined as major adverse cardiovascular events (MACE) at one-year follow-up, which included all causes of death, MI, and TVR. Death was categorized as either cardiac or non-cardiac related. MI was confirmed according to the criteria of "the third universal definition of myocardial infarction". [11] TVR referred to any percutaneous or surgical revascularization of the previously treated vessel. Stroke was diagnosed by CT scan. Bleeding was classified using the "Bleeding Academic Research Consortium (BARC)" definition, [12] in which BARC 3 grade bleeding was considered as major bleeding. The follow-up checkups were performed by telephone after six months and one year after patients' discharge. Death was confirmed upon issuance of a death certificate from the local police office, whereas MI and TVR were confirmed by a certificate of diagnosis from the hospital where the patient was treated.

Statistical analysis
Continuous variables were shown as mean value ± standard deviation and were compared with Student's t test if we assume normal random distribution. Variables that were not normally distributed were shown as medians and quartile ranges and compared using the Mann-Whitney U test. Categorical variables were shown as frequencies and compared using the Chi-square test or Fisher's exact test. Cumulative incidence of one-year MACE was estimated by the Kaplan-Meier curves and assessed using log-rank test. Multivariable regression analysis was performed to identify the independent effect of TRA vs. TFA on in-hospital NACE and oneyear MACE. Variables included in the model were as follows: gender, age, prior MI, prior PCI, prior coronary artery bypass surgery, prior stroke, diabetes mellitus, hypertension, hyperlipidemia, diagnosis of acute coronary syndrome, left ventricle ejection fraction (LVEF), use of GP IIb/IIIa inhibitor, size of sheath, three-vessel disease, left main disease, type C lesion, preprocedure thrombolysis in MI (TIMI) flow, use of drug-eluting stent, and operators with different PCI quantities.
Because the route selection was not randomized, a 1:1 match propensity score analysis was performed to minimize bias. Patients in the TRA group were arranged in order and matched to the most relevant patients in the TFA group by estimating propensity score. Logistic regression analysis was performed to calculate the probability of assignment, and propensity score was calculated from the logistic regression coefficients. Variables included in the logistic model were as follows: gender, age, prior MI, prior PCI, prior coronary artery bypass surgery, prior stroke, diabetes mellitus, hypertension, hyperlipidemia, clinical diagnosis, LVEF, use of GP IIb/ IIIa inhibitor, three-vessel disease, left main disease, post-procedure TIMI flows, use of drugeluting stent, and operators with different PCI quantities. A difference of 0.01 in the estimated propensity score between TRA and TFA indicated that the two patients were characteristically even and were paired together. The discrimination and precision of the propensity score model was evaluated by the area under the receiver operating curve (ROC) in the logistic regression model and the Hosmer-Lemeshow test.
Among the propensity score-matched patients, a separate regression analysis was performed. Independent analyses were also performed within six pre-specified subgroups. All analyses were conducted using SAS software, version 9.13 (SAS Institute, Cary, North Carolina), and a two-sided p value of <0.05 was considered for statistical significance.

Baseline characteristics
A total of 21,242 patients met the inclusion criteria and were included in the analysis. Patients were analyzed based on intension-to-treat principle. From the total, 18,234 patients (85.8%) had PCI from the radial route, and the rest (14.2%) had TFA PCI. The prevalence of TRA in PCI in our institution is increasing. After the propensity score matching, 1,634 pairs were matched. Logistic model showed good stability and showed no discrimination for the matching (ROC area: 0.695; Hosmer-Lemeshow test, P = 0.162). Patients in the TRA and TFA groups were followed up for a median term of 385 d. The follow-up at one year was completed in 17,961 (98.5%) in the TRA group, and 2953 (98.2%) in the TFA group.
Patients' clinical characteristics are shown in Table 1, and the angiographic results are shown in Table 2. Significant differences were detected between TRA and TFA in terms of patients' medical history and severity of atherosclerosis. After propensity score matching, patients included in the TRA and TFA groups had comparable baseline characteristics. However, patients in the TFA group had more lesions treated than those in the TRA group. The size of sheath of more than 7 French was used in 10.3% in the TFA group and 1.8% in the TRA group (P < 0.001), and IABP was more often used in the TFA group (P < 0.001). Catheter-related dissection (1.3% vs. 1.2%, P = 0.931) and thrombosis (0.3% vs. 0.4%, P = 0.897) were similar between the TFA and the TRA group. Contrast volume was higher (152.70 ± 76.79 ml vs. 163.72 ± 95.21 ml, P < 0.001), and the total procedure time was longer (38.59 ± 20.12 min vs. 42.73 ± 25.67 min, P < 0.001) in the TFA group.

Major outcomes
Major outcomes for patients in the TRA and TFA groups are shown in Table 3. The incidence rates of in-hospital NACE (1.8% vs. 4.0%, P < 0.001) and one-year MACE (3.9% vs. 5.2%, P = 0.001) were higher in the TFA group than in the TRA group. However, in the propensity score-matched patients, the rate of in-hospital NACE (1.8% vs. 3.9%, P < 0.001) was higher in the TFA group, and the rate of one-year MACE was similar between TRA and TFA groups (4.7% vs. 4.9%, P = 0.272). The higher rate of NACE in the TFA group was mainly due to the higher rate of major bleeding (0.6% vs. 1.8%, P < 0.001) and the increased rate of MI (1.1% vs. 1.9%, P = 0.060). Meanwhile, in-hospital composite endpoints, including death, MI, and stroke, occurred more in the TRA group (1.2% vs. 2.3%, P = 0.016). The Kaplan-Meier curves of oneyear major outcomes are shown in Fig 2. No statistical difference was detected in MACE or each component of MACE between the TRA and TFA groups in the log-rank test. The total hospital and the post-procedure stays were both longer in the TFA group than in the TRA group (P < 0.001).
The adjusted rates of in-hospital NACE was lower in the TRA group [odds ratio (OR), 0.53; 95% CI from 0.40 to 0.71], and the adjusted rate of one-year MACE were similar between TRA and TFA [hazards ratio (HR), 1.01; 95% confidence interval (CI), from 0.96 to 1.06]. In the propensity score-matched patients, the adjusted rate of in-hospital NACE was still lower in the TRA group (OR, 0.46; 95% CI, from 0.29 to 0.73), and the adjusted rate of one-year MACE remained similar between TRA and TFA (HR, 1.01; 95% CI, from 0.94 to 1.08) ( Table 4).

Access complications and major bleeding
Access site complications and peri-procedure bleeding occurred more in the TFA group (P < 0.05) than in the TRA group. In the propensity score-matched patients, TRA was associated with lower rates of access site complications (1.5% vs. 4.7%, P < 0.001) and access site-related major bleeding (0.4% vs. 1.2%, P = 0.005). However, no difference was detected with regard to non-access site-related major bleeding (0.2% vs. 0.6%, P = 0.076) ( Table 5).

Subgroup analyses
The adjusted rates of in-hospital NACE and one-year MACE in the subgroup patients are shown in S1 Fig and S2 Fig Consistent with the results from the entire patient population,

Discussion
This study compared the outcomes between different route selections in an extremely highvolume PCI institution. Compared with patients in the TFA group, patients in the TRA group had a lower rate of in-hospital NACE, which was mainly due to lower rates of MI and major bleeding. The annual PCI volume of the institution was shown to be a predictor for the TRA outcomes. [10] Because TRA is a skill-demand technique, and centers with high volume PCI are more willing to perform new strategies. TRA implementation is more prevalent in large heart centers. We selected one high-volume PCI center, and results showed that TRA was much safer compared with TFA based on in-hospital outcomes, including lower rates of NACE, major bleeding, and composite endpoints of death, MI, and stroke. This finding was in agreement with the results of previous studies performed in different populations. The RIVAL study showed a lower rate of in-hospital NACE (including death, MI, stroke, and ACUITY bleeding) in acute coronary syndrome patients when TRA was used instead of TFA. [10] The PREVAIL study showed similar rate of short-term death, but lower rate of MI in TRA patients among unselected PCIs. [13] Similarly, we found a decreased rate of MI (1.1% vs. 1.9%, P = 0.060) with TRA instead of TFA, even though it has not reached significance. The underlying reason for the decreased MI in the TRA group is unclear. However, the relatively larger sheath and  guiding catheters in the TFA group could result in more PCI-related MIs than that in the TRA group. The medium-term outcomes between TRA and TFA groups varied across studies. Our study was performed in a stable population undergoing elective PCI, and showed similar rates for one-year MACE between the TRA and TFA groups. The CREDO-Kyoto study showed similar three-year death rates between TRA and TFA in a non-acute MI population. [14] Meanwhile, propensity score-matched Italian cohort with acute MI patients showed lower two-year death rates with TRA than with TFA. [15] Randomized trial results suggested that high-risk patients showed lower death rates with TRA than with TFA. [16][17][18] However, this finding was not detected in any subgroup analysis in the present study.
RIVAL study found lower ACUITY bleeding rate in the TRA group, [10] which was attributed to low rates of access site-related bleeding. Access site-related bleeding accounted for 50% to 70% of all bleeding incidents in patients who underwent PCI. [19] Thus, using TRA instead of TFA could be a good strategy for reducing bleeding events, particularly access site-related bleeding. Access site complications occurred more frequently and were more severe in the TFA group than in the TRA group. The incidence rate of hematomas, which prolonged hospital stay, nearly tripled in the TFA group compared with the TRA group. BARC 2 bleeding (excessive bleeding), which was closely related to access site complications, also occurred more frequently in the TFA group than in the TRA group. Severe complications, such as pseudoaneurysm, arterio-venous fistula, retroperitoneal hematoma are rarely occurred in the TRA group. By contrast, these were frequent complications in the TFA group. Intrinsic anatomical differences exist between the radial and the femoral arteries, thus interventionists can simply accomplish puncturing and easily achieve hemostasis with TRA instead of TFA. [20] In addition, inadequate compression of the femoral artery is the most common reason for the large hematomas. [21] Our study has the following limitations. First, patient assignment was not randomized, but was based on doctor's preference. This could lead to uneven baseline characteristics. However, we used the propensity score to adjust for the disparities. Whereas hidden confounders could not be completely removed, and thus, large randomized trials are required. Second, because of the retrospective extraction, we failed to obtain the crossover rate when one route for the catheterization failed. In analyzing the route effect on major outcomes, we used the intention-totreat principle in dividing patients into groups, which was consistent with the statistical strategy used in previous observational studies. [21] Third, this study was performed in a single heart center with high-volume PCI. Thus, the effect of TRA in high volume PCI centers may be wellrepresented. However, we failed to show the effect of TRA in relatively low or median volume PCI-capable hospitals. Besides, this study included patients from 2006 to 2011, while only 85% of patients had TRA PCI. Nowadays, in Europe at least, centers with high expertise in TRA perform more than 95% of cases by TRA, so the present result might be used as reference in those countries, and the results of those centers are also expected.

Conclusions
Compared with TFA, TRA is much safer and results in lower rate of NACE during hospitalization in patients undergoing elective PCI. We suggest that TRA should be routinely adopted in PCI-capable hospitals.