LADG indicates laparoscopy-assisted distal gastrectomy; ODG, open distal gastrectomy.
eTable 1. Frequencies and Percentages of Missingness for Each Variable by Treatment Arm
eTable 2. Baseline and Neoadjuvant Chemotherapy Characteristics of the As-Treated Population
eTable 3. Surgical, Pathological, and Postoperative Recovery Characteristics of the Modified Intention-to-Treat Population
eTable 4. Postoperative Complications of the Modified Intention-to-Treat Population
eTable 5. Adjuvant Chemotherapy Characteristics of the Modified Intention-to-Treat Population
Data Sharing Statement
Customize your JAMA Network experience by selecting one or more topics from the list below.
Identify all potential conflicts of interest that might be relevant to your comment.
Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.
Err on the side of full disclosure.
If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.
Not all submitted comments are published. Please see our commenting policy for details.
Li Z, Shan F, Ying X, et al. Assessment of Laparoscopic Distal Gastrectomy After Neoadjuvant Chemotherapy for Locally Advanced Gastric Cancer: A Randomized Clinical Trial. JAMA Surg. 2019;154(12):1093–1101. doi:10.1001/jamasurg.2019.3473
How do the short-term outcomes of laparoscopic distal gastrectomy compare with the outcomes of open distal gastrectomy for locally advanced gastric cancer after neoadjuvant chemotherapy?
In this randomized clinical trial of 96 patients receiving neoadjuvant chemotherapy for locally advanced gastric cancer, the overall postoperative complication rate after laparoscopic distal gastrectomy was significantly lower compared with the rate from open distal gastrectomy. Patients in the laparoscopic group also exhibited reduced postoperative pain and better adjuvant chemotherapy completion compared with those in the open gastrectomy group.
The short-term results of this trial suggest that patients with locally advanced gastric cancer may benefit from laparoscopic distal gastrectomy, as lower postoperative complication rates and better adjuvant chemotherapy tolerance were observed among patients receiving neoadjuvant chemotherapy who underwent this treatment.
Laparoscopic distal gastrectomy and neoadjuvant chemotherapy are increasingly used to treat locally advanced gastric cancer. However, the safety and efficacy of the laparoscopic procedure after neoadjuvant chemotherapy remain unclear.
To evaluate the short-term outcomes of patients with locally advanced gastric cancer who received either laparoscopic distal gastrectomy or open distal gastrectomy.
Design, Setting, and Participants
Between April 23, 2015, and November 16, 2017, a phase 2, open-label, noninferiority randomized clinical trial was conducted at the Gastrointestinal Cancer Center of Peking University Cancer Hospital and Institute in Beijing, China. Patients (n = 96) between 18 and 80 years of age with locally advanced gastric cancer (cT2-4aN+M0) who were receiving neoadjuvant chemotherapy were enrolled and randomized. An as-treated population and a modified intention-to-treat (mITT) population were defined for the data analysis.
Patients were randomized to undergo either laparoscopy-assisted distal gastrectomy (LADG) with D2 lymphadenectomy or open distal gastrectomy (ODG) with D2 lymphadenectomy.
Main Outcomes and Measures
The primary end point was 3-year recurrence-free survival rate. Secondary end points were surgical radicality, 30-day postoperative morbidity and mortality, 2-week postoperative recovery indexes, and adjuvant chemotherapy completion status.
In total, 95 patients were eligible for as-treated analyses (LADG: 45, of whom 13 were female [29%], with a median [interquartile range (IQR)] age of 59 [52-65] years; ODG: 50, of whom 16 were female [32%], with a median [IQR] age of 61 [55-64] years) and mITT analyses (LADG: 47, of whom 14 were female [30%], with a median [IQR] age of 59 [52-65] years; ODG: 48, of whom 15 were female [31%], with a median [IQR] age of 61 [55-64] years). In the as-treated analyses, the LADG group had a significantly lower postoperative complication rate than the ODG group (20% vs 46%; P = .007). The postoperative visual analog scale score for pain was 1.2 units lower on postoperative day 2 only in the LADG group (95% CI, −2.1 to −0.3; P = .008). Patients in the LADG group had better adjuvant chemotherapy completion (adjusted odds ratio, 4.39; 95% CI, 1.63-11.80; P = .003) and were less likely to terminate adjuvant chemotherapy because of adverse effects (10 [22%] vs 21 [42%]; P = .04). The mITT analyses showed similar results to as-treated analyses.
Conclusions and Relevance
This trial found that LADG appears to offer the benefits of better postoperative safety and adjuvant chemotherapy tolerance compared with ODG for patients with locally advanced gastric cancer who received neoadjuvant chemotherapy.
ClinicalTrials.gov identifier: NCT02404753
Locally advanced gastric cancer is still the main challenge in the treatment of gastric cancer. The current therapeutic strategy for locally advanced gastric cancer is multidisciplinary with surgical procedure as the core.1 Since the MAGIC (Medical Research Council Adjuvant Gastric Infusional Chemotherapy) trial, neoadjuvant chemotherapies for gastric cancer have been widely accepted.2 Possible advantages include better tolerance, tumor downstaging, and occult micrometastases elimination to improve the chance of radical resection and cure the cancer.2,3
In the past 2 decades, we have witnessed the transformation of gastric cancer surgical procedure from the traditional open gastrectomy to a minimally invasive operation. Laparoscopic gastrectomy has become the established treatment for early gastric cancer. Three ongoing, multicenter, phase 3 randomized clinical trials (RCTs)—namely, the JLSSG 0901 (Japanese Laparoscopic Surgery Study Group) trial in Japan,4 the KLASS-02 (Korean Laparoscopic Gastrointestinal Surgery Study Group) trial in South Korea, and the CLASS-01 (Chinese Laparoscopic Gastrointestinal Surgery Study) trial in China—are trying to extend the indications to locally advanced gastric cancer. The available evidence from the CLASS-01 trial suggested that laparoscopy-assisted distal gastrectomy (LADG) was safe and provided faster postoperative recovery compared with open distal gastrectomy (ODG) for patients with locally advanced gastric cancer.5
However, the safety and efficacy of the laparoscopic procedure in patients after neoadjuvant chemotherapy are unclear. The loss of normal tissue plane from chemotherapy-induced profibrotic reactions and cytotoxicity poses new technical challenges. Whether less trauma is equivalent to better postoperative safety, chemotherapy completion, and survival benefit remains a key question in clinical practice. We have not found any direct evidence of this problem in the literature. A retrospective study of 44 patients showed that LADG after neoadjuvant chemotherapy may provide comparable safety and short-term oncologic efficacy to ODG.6 Furthermore, RCTs were therefore required to test the safety and efficacy of LADG as an acceptable alternative to ODG.
In accordance with the IDEAL (Idea, Development, Exploration, Assessment, Long-term) Study Framework,7 we performed this pilot RCT (ClinicalTrials.gov identifier: NCT02404753), which provides data for estimating the variables required to design a definitive RCT. The primary objective was to determine the noninferiority of LADG (experimental group) compared with ODG (control group) on 3-year recurrence-free survival for patients with a preoperative clinical stage of locally advanced gastric cancer. Data on the primary outcome are currently being collected. The short-term results are presented here.
This phase 2, open-label, noninferiority RCT was conducted between April 23, 2015, and November 16, 2017, at the Gastrointestinal Cancer Center of Peking University Cancer Hospital and Institute in Beijing, China. The trial protocol was approved by the Peking University Cancer Hospital Ethics Committee, and all participants provided written informed consent. Eligible participants were between 18 and 80 years of age with a Karnofsky Performance Scale score higher than 70 and a histologically proven gastric adenocarcinoma with clinical stage of cT2-4aN+M0 by preoperative evaluation. Detailed inclusion and exclusion criteria were described in the published trial protocol (Supplement 1).8
Three weeks after the last cycle of neoadjuvant chemotherapy, patients with resectable tumors were randomized to undergo either LADG with D2 lymph node dissection or ODG with D2 lymph node dissection in a 1:1 ratio (Figure). The data manager (Y.Z.), who was separate from those who conducted the eligibility assessment and recruitment of patients, performed the randomization with a list of randomly ordered treatment identifiers generated by a permuted block design using SAS, version 9.4 (SAS Institute Inc). The allocation sequence was concealed from the surgeons who enrolled patients until the patients had been formally randomized to their groups. Although it was not feasible to blind the surgeons and participants, the pathologists and radiologists were unaware of the treatment received by the patients.
Neoadjuvant chemotherapy was administered to all eligible patients with 3 cycles (3 weeks per cycle) of intravenous oxaliplatin (130 mg/m2 on day 1 of each cycle) plus oral capecitabine (1000 mg/m2 twice daily on days 1 to 14 of each cycle). All patients started to receive 5 cycles of oxaliplatin plus capecitabine regimens within 6 weeks after the surgical procedure. Dose reductions or interruptions were allowed to help patients cope with serious or life-threatening adverse events. Capecitabine could be continued as monotherapy in case of oxaliplatin-associated neurological adverse events, whereas oxaliplatin monotherapy was not allowed. No patients received other forms of perioperative treatments such as radiotherapy, immunotherapy, or targeted therapy.
A standard gastrectomy with D2 lymphadenectomy (including lymph nodes of numbers 1, 3, 4sb, 4d, 5, 6, 7, 8a, 9, 11p, and 12a) was performed for both the LADG and ODG groups in accordance with the Japanese Gastric Cancer Treatment Guidelines 2014, version 49 and the Japanese Classification of Gastric Carcinoma, 3rd English edition.10 In the ODG group, an incision approximately 20 to 25 cm long was made from the falciform process to the periumbilical area. In the LADG group, one 10-mm trocar for the camera was placed below the umbilicus, three 10-mm ports were inserted in the left upper quadrants 2 cm below the left lower-rib margins and the right and left flank areas, and another 5-mm trocar was placed on the right upper quadrants 2 cm below the right lower-rib margins. In both groups, the reconstruction method was selected from the standard procedures of Billroth I/II or Roux-en-Y, depending on the surgeon’s discretion and preference. During the perioperative period, all patients were managed by a modified clinical pathway based on the 2014 Japanese Gastric Cancer Treatment Guidelines.9
The primary end point was the 3-year recurrence-free survival rate. The short-term clinical outcomes reported here included surgical radicality, surgical complications and mortality, 2-week postoperative recovery indexes, and adjuvant chemotherapy completion status. Postoperative complications were defined as events occurring within 30 days after the procedure, the severity of which was assessed by the Clavien-Dindo classification system.11,12 Surgical mortality was defined as all-cause death occurring within 30 days after the procedure. Postoperative pain intensity was measured within 72 hours after the procedure by the visual analog scale.13,14
Baseline statistical consideration and sample size calculation were described previously.8 Conventional 2-tailed superiority hypothesis tests with α = .05 and with 2-sided 95% CIs were adopted for the analysis of safety end points reported here.
Two populations were defined for the analysis: as-treated and modified intention to treat (mITT). The as-treated population excluded the patients who were randomized but did not receive either treatment, and the patients who switched to the other group were seen as the participants of the after-switch group. The mITT population excluded the patients who were randomized but did not receive any treatment, and all the remaining patients were kept in the same group into which they were originally randomized. The as-treated analysis, which was prespecified in the protocol, was considered as the primary analysis for the safety end points. The mITT analysis was the post hoc analysis. Both populations were used for all statistical analyses.
Bivariate analysis was first conducted to compare all characteristics and short-term end points among groups. Unpaired, 2-tailed t test; Mann-Whitney test; χ2 test; and Fisher exact test were used when appropriate. Multivariable regression adjusted for unbalanced preoperative characteristics was then performed for the safety end points.
Because all analyses were exploratory, no Bonferroni correction on P value was used. If missing data were more than 5%, multiple imputations would be conducted. All analyses were performed with Stata software, version 14 (StataCorp LLC) and Rstudio, version 1.1.419 (RStudio Inc).
In total, 96 enrolled patients were randomized to either the LADG group or the ODG group (Figure). After excluding 1 patient who had a rapid disease progression and did not receive any surgical treatment, 95 patients remained in the as-treated analysis population (LADG: 45, of whom 13 were female [29%], with a median [interquartile range (IQR)] age of 59 [52-65] years; ODG: 50, of whom 16 were female [32%], with a median [IQR] age of 61 [55-64] years) and mITT analysis population (LADG: 47, of whom 14 were female [30%], with a median [IQR] age of 59 [52-65] years; ODG: 48, of whom 15 were female [31%], with a median [IQR] age of 61 [55-64] years). Two patients (4%) in the LADG group with open conversion were counted as members of the ODG group for the as-treated analysis. The amount of missing data was less than 5% for important characteristics; therefore, no multiple imputations were performed (eTable 1 in Supplement 2).
The baseline and neoadjuvant chemotherapy characteristics of the mITT population are shown in Table 1, and characteristics of the as-treated population are shown in eTable 2 in Supplement 2. In the mITT population, approximately one-third of participants in both groups were female (LADG: 30%; ODG: 33%). The median (IQR) age distribution of both groups was slightly skewed to the younger population (LADG: 59 [52-65] years; ODG: 61 [55-64] years). The median (IQR) body mass index, calculated as weight in kilograms divided by height in meters squared, was 23.5 (20.9-25.0) in the LADG group and 22.6 (21.2-24.8) in the ODG group. In both analysis populations (LADG vs ODG), statistically significant differences were found in the distributions of computed tomography clinical T stage (T2: 2 [4%] vs 0 [0%]; T3: 19 [40%] vs 31 [65%]; T4a: 26 [55%] vs 16 [33%]; and T4b: 0 [0%] vs 1 [2%]); P = .02), N stage (N0: 10 [21%] vs 22 [46%]; N1: 22 [47%] vs 16 [33%]; N2: 14 [30%] vs 7 [15%]; and N3: 1 [2%] vs 3 [6%]; P = .03), and TNM stage (IIA: 8 [17%] vs 18 [38%]; IIB: 12 [26%] vs 14 [29%]; IIIA: 17 [36%] vs 8 [17%]; IIIB: 9 [19%] vs 5 [10%]; and IIIC: 1 [2%] vs 3 [6%]; P = .048), with the LADG group containing more advanced patients.
All patients started neoadjuvant chemotherapy. Characteristics of neoadjuvant chemotherapy were well balanced between LADG and ODG groups. The exception was found in the as-treated analysis population, in which the LADG group had a significant longer mean (SD) chemotherapy–surgical procedure interval compared with the ODG group (6.2 [1.7] weeks vs 5.6 [1.2] weeks; P = .04; eTable 2 in Supplement 2).
Surgical characteristics of the as-treated population are shown in Table 2. Distal gastrectomy was performed for 43 patients (96%) in the LADG group and 49 patients (98%) in the ODG group (P = .60). All patients had D2 or D2-plus lymph node dissection and achieved negative margins. All patients had R0 dissection except for 1 (2%) in the LADG group and 4 (8%) in the ODG group (P = .37). The LADG group was statistically significantly associated with more Billroth-II with Braun construction (11 [24%] vs 4 [8%]; P = .02) and longer mean (SD) operative time compared with the ODG group (224.8 [35.8] minutes vs 182.9 [44.8] minutes; P < .001). No significant difference was found between the 2 groups regarding median (IQR) intraoperative blood loss (87 [60-150] mL vs 100 [58-200] mL; P = .22) and number of retrieved lymph nodes (31 [24- 38] vs 33 [28- 41]; P = .43).
During the procedure, 2 patients (4%) in the LADG group were converted to ODG because of symptoms suggestive of pancreas invasion. One patient, with pancreatic involvement confirmed by intraoperative pathological findings, underwent distal gastrectomy combined with partial resection of the pancreas and then experienced a grade 2 pulmonary infection after the operation. Meanwhile, the other patient, without pancreatic involvement, recovered without complications.
No statistically significant differences between the 2 groups were found with respect to tumor locations (lower one-third: 33 [73%] vs 40 [80%]; middle one-third: 12 [27%] vs 10 [20%]; P = .44), median (IQR) tumor diameter (2.5 [1.9-3.6] cm vs 2.5 [1.5-3.5] cm; P = .70), and median (IQR) margin length (proximal: 4.5 [3.5-5.5] cm vs 5 [4-6] cm; P = .10; distal: 3 [2.5-5] cm vs 3.5 [2.5-4.5] cm; P = .82) as well as the number of metastatic lymph nodes, tumor regression grade, ypT stage, and ypTNM stage (Table 2). However, in the as-treated population, the LADG group had significantly later ypN stage compared with the ODG group (ypN0: 23 [51%] vs 33 [66%]; ypN1: 14 [31%] vs 6 [12%]; ypN2: 7 [16%] vs 4 [8%]; and ypN3: 1 [2%] vs 7 [14%]; P = .02).
Using the as-treated population, no statistically significant difference was found between the 2 groups regarding mean (SD) time to first aerofluxus (3.3 [0.9] days vs 3.2 [1.2] days; P = .62), median (IQR) time to first defecation (4 [4-5] days vs 4 [4-5] days; P = .91), median (IQR) time to first liquid diet (3 [1-3] days vs 3 [2-4] days; P = .08), and median (IQR) time to first half-liquid diet (7 [6-8] days vs 7 [5-8] days; P = .83) (Table 2). The median (IQR) visual analog scale score was significantly lower in the LADG than in the ODG group on postoperative day 2 (2 [1-3.5] vs 3 [2-5]; P = .04) and day 3 (1.5 [0.5-3] vs 3 [1-4]; P = .04) but not on day 1 (2 [1-3] vs 3 [2-4]; P = .20). After adjustment for unbalanced preoperative characteristics, including the cTNM stage, neoadjuvant chemotherapy–surgical procedure interval, and ypN stage, the visual analog scale score in the LADG group was 1.2 units lower compared with the score in the ODG group on postoperative day 2 (95% CI, −2.1 to −0.3; linear regression P = .008) but not on postoperative day 1 (−0.3; 95% CI, −1.2 to 0.7; linear regression P = .58) and day 3 (−0.4; 95% CI, −1.3 to 0.4; linear regression P = .32).
The median (IQR) number of supplementary patient-controlled analgesia was 0 (0-4) in the LADG group and 2 (0-6) in the ODG group, with no statistically significant difference (P = .14). The median (IQR) postoperative length of hospital stay was 9 (8-10) days in the LADG group and 9 (8-13) days in the ODG group (P = .10). After exclusion of patients with complications of Clavien-Dindo grade 3 and above (n = 7), the LADG group had an approximately 2-day shorter postoperative hospital stay compared with the ODG group (8.6 days vs 10.5 days; P = .004).
Table 3 shows postoperative complications. In the as-treated analysis, the overall complication rate within 30 postoperative days was statistically significantly lower in the LADG group compared with the ODG group (20% vs 46%; P = .007). This protective effect persisted after adjustment for cTNM stage, neoadjuvant chemotherapy–surgical procedure interval, and ypN stage (odds ratio [OR], 0.15; 95% CI, 0.04-0.48; binary logistic regression P = .001). In the comparisons for different complication grades, the LADG group was less likely to have Clavien-Dindo grade II complications compared with the ODG group (6 [13%] vs 20 [40%]; P = .004). Six patients (13%) in the LADG group and 2 patients (4%) in the ODG group had complications of grade III or higher; however, the difference was not statistically significant (P = .25). No unplanned reoperation or death within 30 days after the procedure occurred in either group.
Characteristics of adjuvant chemotherapy in the as-treated population are shown in Table 4. The median (IQR) number of completed cycles was 5 (3-5) in the LADG group and 4 (2-5) in the ODG group (P = .06). After adjustment for cTNM stage, neoadjuvant chemotherapy–surgical procedure interval, and ypN stage, the odds of completing 1 more cycle in the LADG group were 4.39 (adjusted OR) times the odds in the ODG group (95% CI, 1.63-11.80; ordinal logistic regression P = .003). No statistically significant difference was observed with respect to the completion status of each cycle. Patients in the LADG group were less likely than patients in the ODG group to terminate adjuvant chemotherapy owing to adverse effects (10 [22%] vs 21 [42%]; P = .04).
For all of these bivariate analyses, the mITT analysis yielded similar results as the as-treated analysis (eTables 3, 4, and 5 in Supplement 2).
The short-term results of this RCT have shown that LADG can be safely performed by experienced surgeons in patients who have received neoadjuvant chemotherapy and has the benefits of a lower postoperative complication rate and better adjuvant chemotherapy tolerance compared with ODG.
For patients with locally advanced gastric cancer without neoadjuvant chemotherapy, RCTs have consistently found no significant difference in the postoperative complication rate between the LADG and ODG groups.5,15-18 The most recent meta-analysis included 13 RCTs or observational studies and likewise yielded a nonsignificant pooled relative risk of 0.966 (95% CI, 0.77-1.22; P = .77) when comparing laparoscopic with open gastrectomy, but the between-study heterogeneity was large (I2 = 68.3%) and statistically significant.19 In this current study, LADG was found to reduce the postoperative complication rate by half compared with ODG. Therefore, we speculate that the advantages of LADG over ODG on postoperative complications may be more substantial in the neoadjuvant chemotherapy setting, with the following probable mechanism. Chemotherapy can cause tissue fibrotic changes and destruct the anatomical dissection plane, which may contribute to an increased risk of surgical complications.20 However, these issues may be mitigated or even overcome by laparoscopy, which allows visual magnification, better exposure, and more delicate maneuvers of organs, vessels, and nerves. The verification of such a hypothesis, however, requires a formal interaction test enabled by a 2-by-2 factorial RCT design that takes into account both surgical approach (LADG vs ODG) and chemotherapy modality (neoadjuvant vs adjuvant chemotherapy).
Adjuvant chemotherapy is the standard treatment for advanced gastric cancer. It has been commonly observed that laparoscopic operation is associated with better postoperative recovery compared with open procedure, but whether this advantage is translated to better compliance with adjuvant chemotherapy is far less reported for gastric cancer.21 This study found that, compared with the ODG group, patients who underwent LADG were more likely to complete more cycles of postoperative chemotherapy and less likely to terminate chemotherapy because of adverse effects. A similar benefit, that the laparoscopic procedure improved patient compliance with postoperative chemotherapy, was observed in colon cancer.22 However, whether this improved postoperative chemotherapy compliance in the LADG group will translate to better survival is yet to be answered for gastric cancer.
Previous randomized and nonrandomized studies have somewhat consistently reported that the laparoscopic operation compared with open procedure is associated with faster bowel recovery (eg, time to first aerofluxus, time to first defecation, time to first liquid diet, and time to first half-liquid diet) and shorter hospital stay among patients with early or advanced gastric cancer and without neoadjuvant chemotherapy.19,23,24 In the current study, no significant difference in these characteristics was found between the 2 groups. One potential explanation for the null results may be the lack of power given that this trial was not designed to have sufficient power to detect significant difference in these outcomes. Another explanation may be the bias introduced by the unblinded nature of the surgeons. Because no high-level evidence was available on the safety of performing LADG on patients receiving neoadjuvant chemotherapy, surgeons might be more conservative and discreet with the clinical decisions on the postoperative diet and discharge of patients, which would bias the difference toward the null presuming the LADG group was superior.
This study has 2 major limitations. First, only distal gastrectomy was examined, the results of which may be less pertinent to the western health care settings in which proximal gastric cancer is more prevalent and total gastrectomy is required. Second, all surgical procedures in this RCT were performed by 1 senior surgeon (Z.L.) in a single center. This surgeon has performed more than 600 laparoscopic gastrectomy procedures in total and has an annual surgical volume of approximately 300 (including both laparoscopic and open gastrectomy). Therefore, the generalizability of the findings to surgeons with different levels of experience may be limited.
The results of this study suggest that, for patients with locally advanced gastric cancer who received neoadjuvant chemotherapy, LADG can be safely performed by experienced surgeons and has the benefits of a lower postoperative complication rate and better adjuvant chemotherapy tolerance compared with ODG. An RCT with a 2-by-2 factorial design may be needed to examine whether the advantages of LADG on postoperative complications are magnified by neoadjuvant chemotherapy.
Accepted for Publication: June 16, 2019.
Published Online: September 25, 2019. doi:10.1001/jamasurg.2019.3473
Open Access: This article is published under the JN-OA license and is free to read on the day of publication.
Corresponding Author: Jiafu Ji, MD, PhD, Gastrointestinal Cancer Center, Key Laboratory of Carcinogenesis and Translational Research (Ministry of Education), Peking University Cancer Hospital and Institute, No. 52 Fu-Cheng Rd, Hai-Dian District, Beijing 100142, China (email@example.com).
Author Contributions: Drs Ji and Li had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Drs Li and Shan and Ms Ying contributed equally to this work.
Concept and design: Li, Shan, E, Wang, Ji.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Li, Shan, Ying, Zhang, E, Wang, Ren, Ji.
Critical revision of the manuscript for important intellectual content: Li, Shan, Ying, E, Su, Ji.
Statistical analysis: Ying, E, Ji.
Obtained funding: Li, Shan, Wang, Ji.
Administrative, technical, or material support: Li, Shan, Zhang, Wang, Ren, Ji.
Supervision: Li, Shan, Ren, Ji.
Conflict of Interest Disclosures: None reported.
Funding/Support: This study was funded by grant Z151100004015070 from the Beijing Municipal Science and Technology Commission.
Role of the Funder/Sponsor: The funder had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Data Sharing Statement: See Supplement 3.
Additional Contributions: Roberta W. Scherer, PhD, Johns Hopkins University Bloomberg School of Public Health, edited the manuscript. She received no financial compensation for her contribution.