Introduction

Cervical cancer is the third most common malignancy and the second most common cause of death from cancer among women in the USA1. The incidence and mortality of cervical cancer have a large geographic variation as it is more common and more fatal in developing countries2. Cervical cancer is classified into early-stage and advanced-stage cancer3. Although staged clinically, the surgical treatment of early-stage cervical treatment is critical for optimizing patient survival4,5. Radical hysterectomy with pelvic lymphadenectomy is a commonly performed procedure for the treatment of cervical cancer6. Patients who are found to be candidates for surgical intervention may see 5-year survival rate increased to as high as 87%7. Abdominal, laparoscopic, robotic-assisted laparoscopic and vaginal approaches to radical hysterectomy have all been described and performed by many authors8. Secondary to the risk of spreading cancers, several authors have stated that gynecologic oncology has been slower to adopt minimally invasive techniques than other specialties, sometimes reserving these techniques for only risk reducing procedures9. The first laparoscopic radical hysterectomy was performed in 199310 and since that time the minimally invasive surgery (both laparoscopic and robotic-assisted laparoscopic) have been increasingly used.11,12. Over the last two decades, many studies have compared the survival outcomes and operative morbidity of the minimally invasive and open surgery for the management of cervical cancer13. Open radical hysterectomy shows significant morbidity including bladder dysfunction, blood loss, and complications of blood transfusion14. Many retrospective studies15,16,17,18,19 have shown that laparoscopic surgery causes perioperative complications less than open surgery. Complicating this, a recent randomized clinical multicenter trial proved that minimally invasive surgery was accompanied by a high rate of recurrence and a worse disease-free survival rate than open surgery20. In addition, a recent retrospective study that included 2461 patients with cervical cancer showed that minimally invasive surgery is associated with a higher risk of death than open surgery21. The culmination of these results was a change in the National Comprehensive Cancer Network (NCCN) guidelines with regards to minimally invasive radical hysterectomy22. The current NCCN guidelines and European Society of Gynaecological Oncology (ESGO) guidelines recommended that the standard surgical approach for management of early-stage cervical cancer is abdominal radical hysterectomy23.

In addition to this recent evidence against minimally invasive radical hysterectomy techniques, several new high quality trials regarding LRH and RRH have been published. For this reason, we were motivated to investigate LRH versus ORH while excluding robotic assisted techniques. We conducted this meta-analysis with late breaking, high quality data to again compare the efficacy and safety of laparoscopic radical hysterectomy with that of open radical hysterectomy for the management of cervical cancer. We have meticulously excluded all robotic assisted cases, and attempted to obtain data to use as many high quality trials as possible, while excluding robotic assisted cases.

Method

This meta-analysis was performed following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA)24 and the guidelines reported in the Cochrane Handbook for Systematic Reviews of Interventions25.

Eligibility criteria

The inclusion criteria included studies that met all of the following:

  1. 1.

    Population: women with early stage cervical cancer

  2. 2.

    Intervention: Laparoscopic radical hysterectomy (LRH)

  3. 3.

    Comparator: Open radical hysterectomy (ORH)

  4. 4.

    Included Study Designs: randomized clinical trials, prospective cohort, and retrospective cohort studies.

And included at least one of our primary or secondary outcomes.

Primary outcomes included:

(1.) Operative time, (2.) Estimated blood loss, (3.) Rate of intraoperative complications, (4.) Rate of postoperative complications, (5.) Rate of blood transfusions.

Secondary outcomes included:

(1.) Rate of recurrence, (2.) Postoperative or intraoperative mortality (defined as within 90 days of surgery), (3.) Five-year survival rate, (4.) Disease-free survival rate, (5.) Number of resected lymph nodes.

We made all efforts to obtain all data possible from all included studies in order to include as many high quality trials as possible, but ultimately excluded those trials that failed our inclusion critieria because they did not publish, or despite our best efforts we could not obtain the necessary data to differentiate which cases included robotic assistance and which did not. We excluded all secondary works, such as meta-analyses and reviews, all animal studies, conference abstracts, and studies with incomplete reported data.

Search and study selection

We searched PubMed, Medline, Scopus, Web of Science, ClinicalTrials.Gov and Cochrane CENTRAL databases from database inception until January 2nd 2022 for articles that matched our inclusion criteria.

We used the following search strategy in our search: (laparoscopic OR laparoscopy) AND (open OR abdominal) AND ((cervical cancer) OR (cancer cervix)) AND (hysterectomy). We screened the included articles in three steps. The first step implied importing the results from electronic databases to a Microsoft Excel26 sheet using EndNote Software27. The second step included a manual title and abstract screening of the articles imported to the Excel sheet. The third step was the full-text screening of the included citations from step 2. Additionally, we manually searched the references of the included papers for possible missed studies. Two researchers separately performed the literature search and eligibility match. Disagreements in the included studies was reached by consensus. A third member of the research team was assigned to settle disputes if consensus could not be reached on any study's eligibility, but was ultimately never needed.

Data collection

We collected three categories of data from each included study: the first category is the baseline and demographic characteristics of the included participants, such as the author, year, country, sample size, age, BMI, included stages, follow up period, time period, adenocarcinoma, squamous cell carcinoma, and positive lymph nodes. The second category included the outcomes of analysis, mainly: Operative time, estimated blood loss, Intraoperative complication, Postoperative complication, Blood transfusion rate, Recurrence, mortality, Five-year survival rate, Disease-free survival, and Resected lymph nodes. The third category included data for risk of bias assessment. The process of data collection was done using Microsoft Excel26.

Risk of bias assessment

We used the quality assessment tools from the National Heart, Lung, and Blood Institute (NHLB) to assess the risk of bias of observational studies28. We followed The Grading of Recommendations Assessment, Development and Evaluation (GRADE) Guidelines in assessing the quality of this study. We assessed the risk of bias of included trials using Cochrane’s risk of bias tool29. The tool assesses proper randomization of patients, allocation concealment, and adequate blinding through seven domains. Each domain was judged to be either “low”, “unclear”, or “high” risk of bias. The details of the GRADE assessment of each outcome can be found in Supplementary Table S1.

Analysis

We performed the meta-analysis of this study using Review Manager Software27. Our study included continuous and dichotomous outcomes. We analyzed continuous data using mean difference (MD) and 95% confidence interval (CI), while dichotomous data were analyzed using odds ratio (OR) and 95% CI. The fixed-effects model was used when data were homogeneous, while heterogeneous data were analyzed under a random-effects model. To measure the presence of inconsistency among the studies, we used the I2 and the p-value of the Chi-square tests30. Values of P < 0.1 or I2 > 50% were significant indicators of the presence of heterogeneity. We tried to solve the inconsistency of heterogeneous outcomes using Cochrane’s leave-one-out method30.

Ethics approval and consent to participate

This Manuscript has been reviewed by the institutional IRB board at Marchand Institute and was found to be exempt from IRB review. (January 2022). Data used was exempt from consent to participate or publish secondary to the nature of the study being a systematic review, retrospectively looking at previously published data.

Commitment to diversity

The Marchand Institute remains committed to diversity and tolerance in its research, and actively maintains a workplace free of racism and sexism. Greater than half of the authors for this study are female and many represent diverse backgrounds and under-represented ethnic groups.

Results

Summary of included studies

The literature search results are illustrated in the PRISMA flow diagram in Fig. 1. We included sixty studies that met our eligibility criteria from the different databases11,15,16,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87. We analyzed 42,994 patients with cervical cancer in different stages according to FIGO staging (2009 edition)3. A total of 15,995 patients were allocated to the laparoscopic group, while 26,999 patients were allocated to the laparotomy group. The mean age of patients in the laparoscopic group and laparotomy group was 46.5 and 46.7 years, respectively. A summary of the included studies, the demographic data of patients, the included stages, and follow-up duration are described in detail in Tables 1 and 2.

Figure 1
figure 1

PRISMA flow diagram of our literature search.

Full size image
Table 1 Summary of the included studies.
Full size table
Table 2 Baseline characteristics of participants.
Full size table

Results of risk of bias assessment

The result of the quality assessment yielded an overall moderate risk of bias according to Cochrane`s tool and the mean score for observational studies was 10.57 out of 14 according to NHLB. All clinical trials44,55,56,83 reported proper randomization therefore they were categorized as low risk of bias. All clinical trials were at low risk of bias regarding attrition bias and selective reporting. A detailed illustration of the quality assessment of the included studies is illustrated in Supplementary Table S2.

Analysis of outcomes

Operative time (minutes)

Operating time was reported by 42 studies. We found that ORH operating time was significantly lower than LRH operative time (MD = 20.48 [8.62, 32.35], (P = 0.007). Pooled data were heterogeneous (P < 0.001); I2 = 98% which could not be solved by the leave-one-out method or subgroup analysis (Fig. 2).

Figure 2
figure 2

Forest plot of operative time.

Full size image

Estimated blood loss (ml)

We analyzed 6410 patients from 39 studies that reported the estimated blood loss. The overall mean difference showed that blood loss was significantly lower in LRH group than ORH group (MD = − 325.55 [− 386.16, − 264.94] (P < 0.001)), Pooled analysis was heterogeneous (P < 0.001); I2 = 97% (Fig. 3).

Figure 3
figure 3

Forest plot of estimated blood loss (EBL).

Full size image

Intraoperative complication

Thirty two studies reported the rate of intraoperative complications. We found no significant difference between both groups (OR = 1.38 [0.94, 2.04], (P = 0.10)). Data was heterogeneous (P < 0.001); I2 = 69% as shown in Fig. 4A. We solved the heterogeneity by excluding Liang et al.87 (P = 0.15); I2 = 21%, the overall odds ratio after solving heterogeneity did not show any significant difference between both groups (OR = 1.14 [0.86, 1.51], (P = 0.37)) (Fig. 4B).

Figure 4
figure 4

Forest plot of the rate of intraoperative complications before (A) and after (B) solving for heterogeneity by excluding the Liang et al. study.

Full size image

Postoperative complication

A total of 33,563 patients were analyzed from 43 studies that reported postoperative complications. The overall odds ratio showed that the LRH group had a postoperative complications rate significantly lower than that of ORH (OR = 0.70 [0.55, 0.90], (P = 0.005)). Pooled data was heterogeneous (P < 0.001); I2 = 78%. We could not solve heterogeneity by the leave-one-out method or subgroup analysis (Fig. 5).

Figure 5
figure 5

Forest plot of the rate of postoperative complications.

Full size image

Length of hospital stay (days)

Forty-one studies reported length of hospital stay. We found that patients in the LRH group stayed at hospital fewer days than patients in the ORH group (MD = − 3.64 [− 4.27, − 3.01], (P < 0.001)). We found heterogeneity which could not be solved by the leave-one-out method or subgroup analysis (Fig. 6).

Figure 6
figure 6

Forest plot of the average length of hospital stay.

Full size image

Resected lymph nodes

Thirty six studies reported the number of resected lymph nodes as an outcome. Pooled analysis showed that the LRH group was associated with fewer resected lymph nodes than the ORH group (MD = − 2.80 [− 4.35, − 1.24], (P = 0.004)). We found heterogeneity among studies (P < 0.001); I2 = 93% (Fig. 7).

Figure 7
figure 7

Forest plot of the number of resected lymph nodes.

Full size image

Five-year Overall Survival

A total of 8610 patients were analyzed from 22 studies. The combined analysis did not show any significant difference between both groups (OR = 1.24 [0.94, 1.64], (P = 0.12)). A little heterogeneity was found among studies (P = 0.03); I2 = 40% (Fig. 8A). We solved the heterogeneity by excluding Corrado et al.59 (P = 0.21); I2 = 19%. The overall analysis after solving heterogeneity also showed no significant difference between both groups (OR = 1.10 [0.87, 1.40], (P = 0.43)) (Fig. 8B).

Figure 8
figure 8

Forest plot of five-year overall survival before (A) and after (B) solving for heterogeneity by excluding the Corrado et al. study.

Full size image

Disease free survival

Twenty seven studies reported DFS. Pooled analysis did not show any difference between both groups (OR = 1.00 [0.80, 1.26], (P = 0.98)). Data was heterogeneous (P = 0.002); I2 = 50% (Fig. 9).

Figure 9
figure 9

Forest plot of the rate of disease free survival.

Full size image

Postoperative or Intraoperative Mortality (Defined as within 90 days postop)

Twenty four studies reported mortality as an outcome. The overall odds ratio showed no significant difference between either group (OR = 0.86 [0.69, 1.06], (P = 0.15). Pooled analysis was homogeneous (P = 0.14); I2 = 24% (Fig. 10).

Figure 10
figure 10

Forest plot of the postoperative or intraoperative mortality rate.

Full size image

Recurrence

A total of 19,610 patients were analyzed from 40 studies. We found no significant difference between the two groups (OR = 1.01 [0.81, 1.25], (P = 0.95). Data was heterogeneous (P < 0.001); I2 = 56% (Fig. 11).

Figure 11
figure 11

Forest plot of recurrence of disease.

Full size image

Blood transfusion rate

The combined analysis of 12,673 patients from 29 studies favored the LRH group significantly (OR = 0.28 [0.14, 0.55], (P = 0.002). Pooled analysis was heterogeneous (P < 0.001); I2 = 96% (Fig. 12).

Figure 12
figure 12

Forest plot of the rate of blood transfusion.

Full size image

A funnel plot and Egger test of all outcomes can be found in Supplementary File S1. A separate sensitivity analysis involving only the studies with a low risk of bias can also be found in Supplementary File S2. The sensitivity analysis showed different findings only in the outcome of postoperative complications, where the result changed from statistically significant to insignificant. There were no changes in any other outcomes within the sensitivity analysis.

Discussion

This is the largest scale meta-analysis to date that has compared LRH to ORH through an evaluation of the available evidence, and the first, to our knowledge, to attempt to do so while specifically excluding robotic assisted techniques. In our study, we found that LRH was associated with a significantly lower incidence of estimated blood loss, postoperative complications. In addition, predictably, the LRH group had a short period of postoperative hospital stay. The number of the resected in the LRH group lymph node was significantly fewer than that of the ORH group. The combined analysis did not show any significant difference between both groups regarding 5-year overall survival, disease-free survival, and postoperative or intraoperative mortality. The duration of the surgical procedure was significantly longer in the LRH compared to ORH. This stands in contrast to previous studies that showed decreased survival as a result of minimally invasive radical hysterectomy techniques20.

Although isolated studies have disagreed with these results, the majority of previous systematic reviews and meta-analyses have not. In 2015, Wang et al.88 performed a meta-analysis of 12 cohort studies. They demonstrated similar results to our study. They conclude that LRH was better than ORH regarding the short-term outcomes such as faster functional recovery, estimated blood loss, and postoperative complications. The survival outcomes were also similar in both groups.

Smith et al.89 performed another meta-analysis to compare the minimally invasive hysterectomy with abdominal radical hysterectomy. They found that in more than 22,000 women with early-stage cervical cancer the progression-free survival was significantly worse for women who underwent the minimally invasive radical hysterectomy. This finding is strengthened with longer follow-up. The pooled relative risk for postoperative or intraoperative mortality was lower in the minimally invasive group.

In 2020, Kampers et al.90 performed the first meta-analysis stratifying the patients subjected to different operation techniques according to their risk factors. They compared the survival rates of open hysterectomy and laparoscopic hysterectomy in different risk groups. The study demonstrated that protective techniques in laparoscopy result in improved survival.

Kong et al.84 conducted a retrospective analysis of 88 patients with a cervical cancer diameter of 3 cm or greater. They found that LRH can be a feasible alternative surgical procedure for the management of FIGO stage IB and IIA cervical cancer. However, the institution at which the study was performed was introduced with the laparoscopic approach much later than the laparotomy approach. This in turn made the follow-up period in the LRH group relatively short to make any conclusive remarks on survival benefits.

Liang et al.87 reviewed the records of 18,447 patients undergoing radical hysterectomy for cervical cancer. They demonstrated that laparoscopic hysterectomy was associated with a higher risk of major complications than conventional laparotomy. Prior to our review, this study probably represented the largest scale evidence that assessed surgical complications after radical hysterectomy. Other studies to attempt this were single-center studies with small sample sizes15,60,74. In reconciling why the findings of Liang et al. did not also appear as statistically significant in our analysis, we hypothesize that this had to do with limitations on Liang’s study design. This study relied on inpatient medical records or readmission records to obtain information on complications without regular follow-up of the patients. Therefore, without any first-hand knowledge of the circumstances of this study we would hypothesize that perhaps this their conclusion may not have been as statistically significant if direct outreach to patients had been attempted, consistent with other cohort analyses.

In 2021, Campos et al.55 performed a single-center randomized controlled trial on 30 patients with cervical cancer and lymphovascular invasion. They demonstrated a non-significant trend of worse outcomes for LRH. The overall survival time and disease-free survival time were longer in the LRH group. However, the main limitation facing the study was the small sample size.

In 2018, the LACC study by Ramirez et al.20 found that the minimally invasive radical hysterectomy had lower rates of overall survival and disease-free survival than abdominal radical hysterectomy. It was largely based on this study that the National Comprehensive Cancer Network (NCCN) recommended careful counseling of the patient about short-term versus long-term outcomes and oncologic risks of the different surgical approaches.

Limitations

Although the main limitation facing us in this study is the heterogeneity in some outcomes, we managed to understand most of the attributing factors. We believe the use of studies with different designs and disparity in follow-up periods was responsible for the heterogeneity, which is understandable. Another possible limitation of our study includes selection bias, which is difficult to account for. With observational studies, there is always the possibility that surgeons have intentionally selected out cases for laparoscopic radical approaches that appear easier or have different characteristics than those chosen to be performed open. This could affect data. Our analysis represents the most recent and wide-scale evidence that compares LRH to ORH among women with cervical cancer. In light of this controversy it would be desirable to have more well designed prospective randomized trials in order to strengthen the evidence and dissolve any remaining controversy.

Conclusion

We conclude that LRH is associated with a significantly lower incidence of estimated blood loss, postoperative complications. In addition, the LRH had a short period of postoperative hospital stay. In contrast to previous studies that mixed robotic assisted cases we found no difference regarding 5-year overall survival and recurrence excluding robotic assisted cases.