A systematic review and meta-analysis

Background: In this analysis, we aimed to systematically compare the procedural and post-operative complications (POC) associated with laparoscopic versus open abdominal surgery for right-sided colonic cancer resection. Methods: We searched MEDLINE, http://www.ClinicalTrials.gov, EMBASE, Web of Science, Cochrane Central, and Google scholar for English studies comparing the POC in patients who underwent laparoscopic versus open surgery (OS) for right colonic cancer. Data were assessed by the Cochrane-based RevMan 5.4 software (The Cochrane Community, London, UK). Mean difference (MD) with 95% confidence intervals (CIs) were used to represent the results for continuous variables, whereas risk ratios (RR) with 95% CIs were used for dichotomous data. Results:Twenty-six studies involving a total number of 3410 participants with right colonic carcinomawere included in this analysis. One thousand five hundred and fifteen participants were assigned to undergo invasive laparoscopic surgery whereas 1895 participants were assigned to the open abdominal surgery. Our results showed that the open resection was associated with a shorter length of surgery (MD: 48.63, 95% CI: 30.15–67.12; P= .00001) whereas laparoscopic intervention was associated with a shorter hospital stay [MD (–3.09), 95% CI [–5.82 to (–0.37)]; P= .03]. In addition, POC such as anastomotic leak (RR: 0.96, 95% CI: 0.60– 1.55; P= .88), abdominal abscess (RR: 1.13, 95% CI: 0.52–2.49; P= .75), pulmonary embolism (RR: 0.40, 95% CI: 0.09–1.69; P= .21) and deep vein thrombosis (RR: 0.94, 95% CI: 0.39–2.28; P= .89) were not significantly different. Paralytic ileus (RR: 0.87, 95% CI: 0.67–1.11; P= .26), intra-abdominal infection (RR: 0.82, 95% CI: 0.15–4.48; P= .82), pulmonary complications (RR: 0.83, 95% CI: 0.57–1.20; P= .32), cardiac complications (RR: 0.73, 95% CI: 0.42–1.27; P= .27) and urological complications (RR: 0.83, 95% CI: 0.52–1.33; P= .44) were also similarly manifested. Our analysis also showed 30-day re-admission and re-operation, and mortality to be similar between laparoscopic versus OS for right colonic carcinoma resection. However, surgical wound infection (RR: 0.65, 95% CI: 0.50–0.86; P= .002) was significantly higher with the OS. Conclusions: In conclusion, laparoscopic surgery was almost comparable to OS in terms of post-operative outcomes for rightsided colonic cancer resection and was not associated with higher unwanted outcomes. Therefore, laparoscopic intervention should be considered as safe as the open abdominal surgery for right-sided colonic cancer resection, with a decreased hospital stay. Abbreviations: CI = confidence intervals, LS = laparoscopic surgery, MD = mean difference, OS = open surgery, POC = postoperative complications, RR = risk ratios.


Introduction
Dense fibrous connective tissues, or dense regular connective tissues, are predominantly collagenous tissues with dense and regular orientation of the fibers with respect to each other. [1] They are found in highly fibrous tissues such as ligaments, tendons, fascia, and aponeuroses and are known for their high tensile strength. [2] Unlike skeletal muscle and bone, which are to some degree capable of regeneration, dense fibrous connective tissues heal by the formation of collagen and scar tissue after being injured. The healing process is characterized by angiofibroblastic hyperplasia, including hypercellularity, neovascularization, increased protein synthesis, and matrix disorganization. [3][4][5] In general, this process is slow, [6] and the resulting fibroblastic scars often possess inferior mechanical and biochemical properties compared to native tissues. [7,8] These factors can contribute to chronic pain and disabilities observed in patients with tendinopathies, fasciopathies, and ligament injuries. So far, experts have not agreed upon an effective treatment that optimizes the healing process.
Several injective medications have been tried to facilitate the healing process after fibrotic tissue injuries. Corticosteroid injection therapies have been used in managing these injuries; however, the lack of inflammation in the healing process, along with poor long-term outcomes [9,10] and adverse effects, [11][12][13] have led investigators to question the use of corticosteroid injections. Prolotherapy with hypertonic dextrose (PrT) is an available option in clinical practice. Advocates have suggested that such injectates may induce an inflammatory process, initiate the body's wound-healing cascade, and lead to cellular proliferation, collagen deposition, and eventually tissue repair, [14][15][16] thereby leading to pain reduction and functional improvement. Recently, PrT has become increasingly popular in the United States and internationally in managing various soft tissue problems. [17] Several reviews have investigated the effectiveness of PrT for individual pathologies such as temporomandibular joint hypermobility [18] and Achilles tendinopathy, [19] and some network meta-analyses have compared all of the injection therapies, including PrT and corticosteroid injections for rotator cuff tendinopathy [20] and lateral epicondylopathy. [21,22] However, no definite conclusion was drawn due to insufficient high-quality randomized controlled trials (RCTs). Though the abovementioned structures are not histologically identical, they are all made up of dense fibrous connective tissue and share some similarities. For instance, they are composed of abundant parallel-ordered collagen fibers; they rely on the hierarchical structure to resist tension and stretch; and they all have a slow healing process. Pooling studies involving these structures were hence reviewed to provide insights into the effects of PrT on this histologic entity.
Therefore, a systematic review with meta-analysis was conducted to explore both the effectiveness (prolotherapy vs placebo or no treatment) and superiority (prolotherapy vs corticosteroid injection) of PrT regarding pain control and activity improvements in patients with dense fibrous connective tissue injuries.

Methods
This review study was reported in accordance with the PRISMA guidelines and registered with PROSPERO (CRD42019129044).

Eligibility criteria
This study included RCTs published in peer-review journals, and focused on studies which included adult participants diagnosed with dense fibrous connective tissue injuries, including injuries to tendons, ligaments, or fascia, for which they received injection therapy. As prolotherapy may refer to injections of various proliferent agents, this review will limit the scope to hypertonicdextrose injection. Studies were eligible if they compared the treatment effects of PrT with placebo, no PrT, or corticosteroids, and evaluated either pain or the activity level at follow-up. Cointerventions (e.g., physical therapy) were allowed if they were arranged in the same condition for comparing groups. Injections to irrelevant tissues (e.g., intra-articular, intramuscular, subcutaneous, or perineural) were not considered.

Study Identification
Relevant articles were searched in the PubMed, Scopus, and Embase databases from the earliest record to February 18, 2019. Main search terms were "(prolotherapy) OR [(dextrose OR glucose) AND (tendin * OR tendon * OR ligament OR fasci * OR joint * OR arthr * OR epicondyl * )]." (See Supplemental Table I, http://links.lww.com/MD/F206, Supplemental file, which displays our search plan). The Cochrane Library and Google Scholar were scrutinized for additional references. Three authors (MWC, CYH, and WKC) searched and evaluated the literature for inclusion of studies based on their titles and abstracts. After pooling studies obtained from different sources and removing duplicates, the full texts of potentially relevant articles were retrieved, and each article was independently evaluated by MWC, CYH, and WKC for eligibility. The involved articles were exported to EndNote 5.4 (Clarivate Analytics) for review.

Quality assessment
This study assessed the quality of included studies using the Physiotherapy Evidence Database (PEDro) scale. The methodological quality was assessed by ten items regarding random allocation, blinding procedures, and the dropout rate and statistical reporting. Aggregate scores ranged 0 to 10 points with a higher score indicating better quality. Quality was classified as high (6-10), fair (4 or 5), and poor (3). Using the Cochrane risk of bias tool, this study assessed seven domains of bias and stratified the risk of bias into low, high and unclear risk. Discrepancies between reviewers at any stage were resolved through discussion and consensus.

Outcomes
This study investigated the treatment effects on pain reduction and activity improvement. Pain reduction was assessed by the subjective perception of pain severity or satisfaction with the pain condition, including using a visual analogue scale, Likert scale, or any other continuous pain scale. Activity improvement was measured by questionnaires about activities of daily living or disabilities (e.g., Shoulder Pain and Disability Index, Patient-Rated Tennis Elbow Evaluation).

Data extraction
This study extracted relevant data from each study with a standard data recording form. Data of three time points was Chung et al. Medicine (2020) 99: 46 Medicine collected to evaluate the immediate (i.e., 0-1 month after the first injection), short-term (i.e., 1-3 months after the first injection), and long-term (i.e., 6-12 months after the first injection) effects of the interventions. If a study included multiple measures within the above-mentioned intervals, the measurements closest to 0, 3, and 12 months after the first injection were selected as the immediate, short-term and long-term follow-up data, respectively. The means, mean changes, and corresponding standard deviations (SDs) of outcomes in the three follow-up periods were extracted. One study can be used only once in one comparison. If a study used PrT in more than one experimental group, [23,24] an estimated mean SD would be calculated by merging means and SDs from the experimental groups. If a study contained a placebo and no-PrT groups as the control groups, the results from the placebo group was used to assess the effectiveness.

Data analysis
The analyses were performed using Review Manager Software 5.4. Studies comparing PrT to placebo or no PrT were reanalyzed and interpreted individually to understand the effectiveness of PrT on pain control in the short-and long-term at individual studies level. A meta-analysis which pertained to the comparison "PrT vs placebo or no PrT" and "PrT vs corticosteroids" was then conducted separately for the three time points of interest. [25] The meta-analysis aimed to evaluate the overall effectiveness and superiority (compared to corticosteroids) of PrT in respect of pain and activity improvements. Standardized mean differences (SMDs) were obtained to assess the effect size. A random-effects model was used, and a point estimate with a 95% confidence interval (CI) was presented. Heterogeneity across studies was tested using the I 2 test. An I 2 score of >50% indicated significant heterogeneity.

Results
Five hundred seventy non-duplicated records were yielded. After exclusion based on the title, abstract, full-text review, and the same study sample, seven effectiveness [23,24,[26][27][28][29][30] and three superiority [31][32][33] (compared to corticosteroids) studies were included for review. Figure 1 displays the flow diagram of study development. In total, 10 studies regarding rotator cuff tendinopathy (n = 3), lateral epicondylitis (n = 3), temporomandibular joint hypermobility (n = 2), Achilles tendinopathy (n = 1), and plantar fasciitis (n = 1) involving 358 participants were reviewed and analyzed. Table 1 displays the main characteristics of the included studies. Of the seven effectiveness studies, five were placebocontrolled studies, [23,[26][27][28][29] and co-interventions of physical therapy were performed in two studies. [26,30] The number of total injections ranged from one to 12, while the interval ranged from once every week to once every month. In one trial, [30] the number of total injections differed from patient to patient and ranged from four to 12. The follow-up period ranged from 6 weeks to 3 years.
PEDro scores ranged from 5 to 10, with medians of 7.3 for effectiveness studies and 6.3 for superiority studies (compared to corticosteroids). (See Table 1 and Supplemental Table ii, http:// links.lww.com/MD/F206, Supplementary file, which displays the PEDro scale of each study.) Only three trials [28][29][30] reported a suitable method for allocation concealment. Three studies [24,30,33] had high risks of bias in the blinding of participants and personnel. Only three [26,28,31] studies presented a successful method of outcome assessor blinding. Most studies reported an adequate description for incomplete results, generating unclear risk in presenting reporting bias. In general, most of the included studies had low-to-moderate risks of bias. (See Supplemental www.md-journal.com Table 1 Summary of included studies.    Supplemental Table iii, http://links.lww. com/MD/F206, Supplementary file, which displays the studylevel evaluations of each study). Across comparisons in various disorders, all the study results demonstrated non-significant mean differences between the groups. Two comparisons from one Achilles tendon [30] and one rotator cuff [26] study found significant mean change difference indicating that PrT might be effective in improving pain in the long-term. Figure 2 outlines the effectiveness and superiority (compared to corticosteroids) of PrT at different time points. No significant SMD was found regarding its effectiveness on pain control at any time point (i.e., immediate, short-term, long-term) ( Fig. 2A). PrT was only superior to corticosteroids in the short-term (SMD: 0.70; 95% CI: 0.14-1.27; I 2 = 51%) but inferior in the immediate-term, and not superior in the long-term (Fig. 2B). PrT was effective in improving activity only in the immediate-term (SMD: 0.98; 95% CI: 0.40∼1.55; I 2 = 0%) (Fig. 2C), but not superior to corticosteroids at any time point (Fig. 2D) Considering that histological features of peri-temporomandibular joint soft tissues (i.e., synovial capsule) can differ from the other soft tissues of interest (i.e., ligament, tendon, and fascia), a subgroup analysis was performed after removing two studies [23,27] involving the temporomandibular joint. No change of significance of original SMDs in any outcome categories was found. Sensitivity analyses after removing 2 studies [24,30] without placebo control also did not result in significance changes of original SMDs.

Discussion
This review investigated the effects of PrT on various fibrous connective tissue injuries. The majority of included studies were of moderate-to-high quality and possessed minor-to-moderate risks of bias. The results of the analysis at individual study level and the meta-analysis were inconsistent. In general, the majority of the comparisons of did not yield positive results. Consequently, this study suggests that there is insufficient evidence to support the clinical benefits of PrT in managing fibrous tissue injuries. Prolotherapy or proliferative therapy is a treatment option for damaged connective tissues involving the injection of a solution (proliferant) which theoretically causes an initial cell injury and a subsequent "proliferant" process of wound healing via modulation of the inflammatory process. [34] Several in vitro studies have shown that cells exposed to hypertonic glucose have an initially decreased viability in terms of decreased cell counts, DNA synthesis, and cellular metabolic activities, [14,16,35] as well as an inflammatory reaction. [15] However, it is unclear whether the subsequent "proliferant" process can lead to better outcomes.
Freeman et al administered various dosages of P2G (namely phenol, glycerin, and glucose) to mouse preosteoblast cells and patellar tendon fibroblasts in vitro. In their best result, only the group treated with 25 mL/mL P2G was associated with a higher cellular viability of preosteoblasts compared to the control group, which was noted only at weeks 2 to 3 during the 6-week observation period. Also, such superiority was not seen in fibroblast viability or collagen production. [16] Martins et al assessed the histology of collagen fibers after administering prolotherapy with 12.5% dextrose into rat Achilles tendons, and found no changes in neovascularization or fibroblasts numbers. [36] Perhaps the strongest support of prolotherapy came from a Korean language journal. [17] Kim et al reported that chondrocytic tissue filling of 2-mm punch lesions in adult rabbit femoral cartilage was present 6 weeks after injection of 10% dextrose but not after injection of controls. [37] Ahn et al and Kim et al reported that significantly more fibroblasts were recruited after a dextrose injection into injured and non-injured rat Achilles tendons. [38,39] Whether these findings are reproducible and applicable to the human body remains to be seen.
Three of the included trials in this review used imaging methods to assess recovery following prolotherapy. In a trial conducted on patients with lateral epicondylitis, Rabago et al reported no within-or between-group changes in magnetic resonance imaging scores of common extensor tendons despite the better clinical outcomes associated with prolotherapy. [24] Similarly, a trial conducted by Bertrand et al reported no between-group differences in an Ultrasound Shoulder Pathology Rating Scale, while reporting positive effects of prolotherapy on clinical outcomes in patients with rotator cuff tendinopathy. [26] Lin et al also reported no between-group differences in histograms or sonographic morphology in a study involving supraspinatus tendinopathy. [28] In general, histologic and imaging evidence supporting prolotherapy-induced cell proliferation are still lacking, and further studies are required to establish the effects and mechanisms of PrT.
A number of reviews have previously evaluated the effects of prolotherapy on various body parts. A meta-analysis of three temporomandibular joint studies suggested that PrT might lead to significant reductions in mouth opening and associated pain. [18] A network meta-analysis study of rotator cuff tendinopathy including only one prolotherapy trial reported that prolotherapy was effective over 24 weeks. [20] Two network meta-analyses for lateral epicondylitis which respectively included only one and two RCTs suggested that prolotherapy resulted in better outcomes than placebo. [21,22] A meta-analysis study of Achilles tendinopathy including only one RCT stated that eccentric loading exercise combined with prolotherapy provided more-rapid symptomatic improvements than exercise alone in the short term. [19] Given that most of these reviews were based on a limited number of RCTs, they provide very weak evidence as to the effects of prolotherapy.
The present study updated the current knowledge and considered all dense fibrous connective tissue injuries as a whole. However, still only a limited number of high-quality studies explored the beneficial effects of PrT. Considering that prolotherapy is a cheap and convenient treatment option for managing soft-tissue disorders with less probable side effects, more clinical and basic studies are warranted to fully explore its potential benefits.

Limitations
Several limitations should be addressed. The involved study population differed in diagnosis, durations of symptoms, mechanisms, severity of injuries, and methods of injection, which potentially contributed to the evident heterogeneity. Although the tendons, ligaments, and fascia shared many common features, the surrounding environment, vasculature, and tensile loads within these structures vary. Finally, only three databases were searched, and only a limited number of trials and participants were available for this analysis.

Conclusions
There is insufficient evidence to support the clinical benefits of dextrose prolotherapy in managing fibrous tissue injuries, either in aspect of pain management or activity improvement. More high-quality randomized controlled trials are warranted to establish the benefits of prolotherapy.