PubMed, Embase, Cochrane Library, Medline, China National Knowledge Infrastructure (CNKI), Chinese Scientific Journal Database (VIP), Wanfang Database, and Sinomed were systematically searched from inception to 30th June 2022. Search terms included the following: (“Human Papillomavirus” or “HPV Human Papillomavirus” or “HPV Human Papillomaviruses” or “Human Papillomavirus, HPV” or “Human Papillomaviruses, HPV” or “Human Papillomavirus” or “Human Papillomaviruses” or “Papillomavirus, Human” or “Papillomaviruses, Human”) and (“TCM” or “traditional Chinese medicine” or “Chinese medicinal herb” or “Chinese herbal medicine” or “decoction” or “formula” or “prescription” or “Chinese patent medicine” or “Chinese patent drug” or “Chinese herbal compound prescription”) and (“randomized controlled trial” or “controlled clinical trial” or “random” or “randomly” or “randomized” or “control” or “RCT”). The authors of the identified studies were contacted for additional information if necessary.

Randomized controlled trials (RCTs) published in any language were included. The study population (P) referred to women diagnosed with HR-HPV positive cervical lesions diagnosed as a low-grade squamous intra-epithelial lesion (LSIL) or below by histopathological examination, meeting the diagnostic points formulated by ASCCP (10) and the Textbook of the 12th five-year plan of the Ministry of Health, Obstetrics, and Gynecology (People's Medical Publishing House, 8th edition, 2018, edited by Xie Xing and Kong Beihua). The interventions (I) were HDCM used alone, including both single Chinese herbs and Chinese herbal formulae such as decoction, capsule, tablet, granule, or pill. The controls (C) were interferon or follow-up groups. The primary outcomes (O) involved the rate of HR-HPV clearance, the reduction of HPV-DNA virus load, and the regression rate of CIN at different follow-up time points. The secondary outcomes covered the rate of recurrence and adverse events.

(1) Cervical lesions diagnosed as a high-grade squamous intraepithelial lesion (HSIL) or above by histopathological examination; (2) HDCM combined with acupuncture, TCM fumigation therapy, or other Chinese medicines; (3) duplicate literature; (4) animal experiments; (5) research data exposed serious errors or the full text was not available.

Two authors (HS and QYJ) independently extracted the following:

Data: the first author, publication year, characteristics of participants, intervention methods, components of HDCM, type of HPV, duration of treatment, follow-up time, and outcome measures. Any discrepancy was resolved by consultation among the reviewers or discussion with a third author (Jinbang Xu).

The Cochrane Risk of Bias Tool (19) was adopted to evaluate the quality of trials, focusing on seven aspects: (1) random sequence generation; (2) allocation concealment; (3) blinding of participants and personnel; (4) blinding of outcome assessment; (5) incomplete outcome data; (6) selective outcome reporting; and (7) other sources of bias. The evaluation criteria for each item were judged as “low risk of bias”, “unclear risk of bias”, and “high risk of bias”.

Data analysis was completed using Review Manager 5.3. Risk ratio (RR) and mean difference (MD) with their 95% confidence intervals (CI) were employed for dichotomous outcomes and continuous outcomes separately. Heterogeneity was measured by I² statistics. Specifically, a fixed-effect (FE) model was utilized if I² < 50%. Otherwise, the random-effect (RE) model was used. A p ≤ 0.05 was considered statistically significant. Publication bias was explored by means of funnel-plot analysis.

The sources of heterogeneity were analyzed. The robustness of the results was tested by conducting the sensitivity analysis with the exclusion of studies with unclear random sequence generation. The effect of HDCM with different treatment durations was analyzed separately with subgroup analysis.

A total of 880 records (PubMed [n = 72], Cochrane Library [n = 18], Medline [n = 23], Embase [n = 1], CNKI [n = 627], Sinomed [n = 101], VIP [n = 10], and Wanfang Data [n = 28]) were retrieved, with 150 duplicates removed. After reviewing the titles and abstracts, 676 records were excluded, leaving 54 records remaining. 12 RCTs were finally included after full text screening (Figure 1) (20–31): 1 in English and 11 in Chinese.

The enrolled studies published between 2011 and 2021 in China were utilized, including a total of 1,574 patients. Among them, seven studies (20–24, 27, 30) compared the treatment effect of HDCM with interferon, and six (20, 26, 28–31) focused on the comparison of follow-up. Nine studies applied vaginal preparations of HDCM, and three studies (20, 27, 28) used oral HDCM. Only one study (20) reported specific HR-HPV types. The follow-up duration ranged from 1 to 12 months. Table 1 presents the characteristics of the included studies.

Only one study clearly reported random allocation sequence and allocation concealment (21). Two trials (24, 31) were marked as “high risk” since random design was not mentioned. The other eight trials were marked as “unclear risk” because they failed to report details on how to randomize. Apart from one trial (21), none of the other 11 trials elaborated on whether the double-blind mechanism for participants and personnel was implemented or not. Considering the objective outcome indicators used in all of these studies, researchers did not alter the results, and detection bias was identified as “low-risk.” Three trials (23, 29, 31) reported missing data. In addition, all studies reported all outcomes mentioned in part on methods, and selective reporting bias was assessed as “low-risk.” Other biases were defined as “unclear risk” by reason of the limited information provided in related reports. Figure 2 provides a summary of the risks of bias.

As for the sensitivity analysis, each study in the meta-analysis was screened to reflect the impact of individual data on changes in heterogeneity. Li (21) was the main source of high heterogeneity. The heterogeneity may arise from the following aspects: (1) The HDCM components, as well as the dosing scheme of intervention measures among studies, are far removed from each other; (2) Chinese medicine usually takes a period of time to produce curative effects; (3) There are differences in HR-HPV subtypes and course of disease among participants in the study, which may give rise to higher heterogeneity. However, all of the results from the studies showed the advantages of HDCM treatment.

Publication bias analysis was investigated for each group in this study, respectively. The obtained funnel plot was asymmetric, indicating a mild publication bias (Supplementary Figures).

HR-HPV infection is the leading cause of cervical cancer, which may trigger the immune escape of cervical cells, making them enter the immortalized state (32). To our knowledge, this study was the first to conduct a systematic review and meta-analysis of published RCTs to assess the efficacy of HDCM in treating cervical HR-HPV infection. It was revealed that HDCM could significantly enhance the clearance rate of HR-HPV, reduce HPV-DNA virus load, and increase the regression rate of CIN. The pooled analysis indicated that the HPV clearance rate of HDCM groups was 1.4-fold higher relative to interferon groups and 3.2-fold higher relative to follow-up groups. In terms of remission of CIN and HPV-DNA virus load, significantly greater anti-malignancy efficacy was detected in HDCM groups than in follow-up groups (RR = 2.80,95% CI:2.19, 3.57, P < 0.01), as well as evident antiviral efficacy (MD=-5.16, 95% CI: −5.91, −4.41, P < 0.01). Similarly, at the end of follow-up, the recurrence rate of HR-HPV in HDCM groups was reported to be lower than that in follow-up groups. Moreover, only mild and transient adverse events were reported, further suggesting the safety of HDCM.

Traditional Chinese medicine, which has been developed for thousands of years in China, is one of the most vital complementary and alternative medicines. There is a large body of evidence that suggests that HDCM can exert a positive effect on treating HPV infection. The Phellodendron chinense C.K.Schneid. Ketone from Phellodendron chinense var.Glabriu-sculum C.K.Schneid. can increase the expression time of the mitogen-activated protein kinase phosphatase-1(MKP-1) protein by reducing the transcription and translation levels of NO, IL-6, IL-1β, and other inflammatory factors while stabilizing the mRNA of MKP-1 by significantly inhibiting p38-mediated AP-1 signaling (33). Matrine is an important ingredient of Sophora flavescens Aitonis, characterized by its excellent properties against tumors by accelerating cell apoptosis, inhibiting tumor cell growth and proliferation, inducing cell cycle arrest, and preventing cancer metastasis and invasion. Moreover, it can also inhibit angiogenesis, induce autophagy, reverse multidrug resistance, and inhibit cell differentiation (34). 2-Methoxy-6-acetyl-7-methyljuglone (MAM), a natural naphthoquinone found in Reynoutria japonica Houtt., is proven to be effective against the progression of colon carcinoma by inducing necroptosis in cancer cells through JNK activation and mitochondrial ROS production (35). Shikonin, a naphthoquinone extracted from Arnebia hispidissima DC., can kill cancer cells in a necroptosis manner (36). Researchers explored that (37) Hedyotis diffusa Spreng. can activate murine and human antigen-presenting cells via the MAPK signaling pathway while enhancing antigen presentation in bone marrow-derived dendritic cells in vitro. Besides, Rutin, one of its ingredients, can trigger a strong, specific immune response against HPV-related tumors in vivo. Therefore, it can be reasonably speculated that the combination of HDCM can theoretically regulate the immune response, induce cell necroptosis, and enhance anti-inflammatory reactions, eventually showing a significant treatment effect on HR-HPV infection.

Some limitations exposed by this study cannot be ignored. First, all of the studies involved were conducted in China, with only studies published in Chinese and English enrolled. Consequently, it is difficult to determine whether HDCM is effective for different populations worldwide. Second, with the exception of one study (20) published in the English language and available on Pubmed, most of the studies included were published in Chinese. Among the other 11 studies, two of them were available as abstracts in English with no full text, and the remaining studies were published solely in Chinese. As a result, only one of the included RCTs can be critically assessed by readers from other countries outside of China, which is one of the main shortcomings of the article.

Third, the TCM diagnosis of diseases is mainly based on the evaluation of a series of subtle variables and patient characteristics. Doctors generally judge the curative effect by collecting the characteristics of these subtle changes in patients after using drugs, which may cause a higher risk of allocation bias. Therefore, strict allocation and concealment become more important in RCTs evaluating the therapeutic effect of TCM. The descriptions of randomization, allocation concealment, and blinding methods in most of the included studies are insufficient or absent, which leads to the inability to fully and systematically evaluate the internal validity of studies. Fourth, the heterogeneous characteristics of all patients are a significant limitation of our study. Possible confounding factors included patients' age, type of HPV, course of the disease, HDCM components, and dosage regimens. Fifth, the combination, dosage form, and chemical composition of traditional Chinese medicine in the HDCM formula in each study are different, which is also one of the reasons for the heterogeneity of studies. Sixth, the follow-up time of most studies is too short to provide strong evidence for data research and analysis of the recurrence rate. Furthermore, the lack of HPV infection genotype in most studies makes it impossible for this study to obtain multiple infection data from original studies, which also makes it difficult to determine whether it is a recurrence or reinfection.