After data mining the PubMed database, we identified several relevant studies on cervical cancer/cervical carcinoma and genetic variations (polymorphisms/single nucleotide polymorphisms, SNP/SNV/Mutation). To be included in the meta-analysis, studies had to meet certain conditions. These included reporting covariate risk factors such as HPV status, smoking status, oral contraceptive use, age at menarche, age at menopause, age at first pregnancy, and parity. They also had to provide case-control genotype data for the polymorphic variant, complete research articles, and be published between 2015 and 2023. Only polymorphisms referred to in two or more papers were considered.

Six papers were chosen for further meta-analysis, including data on various factors. When citing a study, it is essential to include critical details such as the first author's name, year of publication, median age (including standard deviation), gender distribution, smoking and HPV statuses, ages at menarche and menopause, ages at marriage and first pregnancies, number of children, use of oral contraceptives, genetic variations, and specific data on case-control groups. The selected papers specifically focused on one gene and one genetic polymorphism. The Newcastle-Ottawa Scale was created to evaluate the quality of case-control studies that are deemed acceptable12. This scale comprises three domains: selection, exposure, and comparison, each assessed based on five parameters, including the Hardy-Weinberg equilibrium (HWE) index, the number of cases and controls, the method of association assessment, and the genotyping method. A maximum score of 8 indicates high quality in the evaluation process.

Individual analyses were conducted for gene polymorphisms with at least two available studies, estimating Odds Ratios (OR) and their corresponding confidence intervals (CI). These were derived from the primary data obtained during the review. We estimated odds ratios (ORs) for each genetic polymorphism and conducted heterogeneity tests using Cochran's Q to assess variations within and between studies. The null hypothesis for these tests suggests that any observed variability is due to chance alone, indicating no significant differences in associations among studies. The alternative hypothesis proposes the presence of heterogeneity in the associations between studies. To quantify the level of heterogeneity, we calculated I² based on Cochran's Q. The I² value indicates the degree of heterogeneity, with low heterogeneity (I²: 25-50%), moderate heterogeneity (I²: 50-75%), and high heterogeneity (I² > 75%)13, 14. If I² is less than 50%, we used a fixed-effects model; if I² is more than 50%, random effects were chosen to interpret the results.

Heterogeneity indicates notable variations among the studies included in the meta-analysis. Delving deeper into the origins of this heterogeneity could offer valuable insights. Potential sources of heterogeneity may stem from differences in study design, characteristics of the study populations, or methods of measurement. Exploring these factors could elucidate why certain studies diverge in their findings and enhance the understanding of the overall results.

The Cochran's Q test was employed to determine whether to use the random or fixed-effects models for pooled odds ratio estimation for each polymorphism. The random-effects model was utilized if the two measures showed significance with a significance level (α) of 0.05 (corresponding to at least moderate I²). If both tests were not statistically significant (equivalent to low I²), fixed-effects models were employed. We performed sensitivity analyses on polymorphisms with more than four studies. To ensure consistency of results, this was done by removing studies iteratively and recalculating the pooled odds ratio (OR) each time. Additionally, the Egger Test was used to assess publication bias for polymorphisms with more than five studies, regardless of whether a significant association was observed15, 16.

The significance level for all analyses was set at 0.05, a commonly used value in the field's original studies. This choice was made to address the limitation of statistical power in the meta-analysis and minimize the risk of Type II errors. Despite this, all reported p-values are provided for transparency, allowing readers to interpret the results critically, regardless of whether the associations were significant17, 18.

Figure 1

Figure 2

Figure 3

Table 1

Study Name	Year	Country	Source of DNA	No of Cases	No of Control	NOS Score	Genotyping Method
Agne et al.	2021	Lithuania	Blood	89	84	6	Real-time PCR
Kushwah et al.	2020	India	Venous Blood	246	246	6	PCR-RFLP
Maneesh et al.	2015	India	Peripheral Blood	100	100	7	PCR-RFLP
Monishita et al.	2023	Bangladesh	Venous Blood	126	120	8	PCR-RFLP
Sabrina et al.	2016	Tunisia	Blood	112	164	8	Real-time PCR
Sayma et al.	2020	Bangladesh	Blood	252	228	7	ARMS-PCR

Table 2

Author	Genotype						Allele				X2	HWE
	Case			Control			Case		Control
	GG	GA	AA	GG	GA	AA	G	A	G	A
Agne et al.	25	41	23	27	44	13	98	70		238	0.5	0.47
Kushwah et al.	86	101	59	182	47	17	411	81	445	573	22.9	0.000002
Maneesh et al.	35	40	25	77	20	3	174	26	180	226	1.34	0.24
Monishita et al.	8	68	50	4	19	97	27	213	221	453	5.14	0.02
Sabrina et al.	4	34	74	15	33	116	63	265	295	593	20.2	0.000007
Sayma et al.	14	150	88	3	39	186	45	411	417	867	0.33	0.56

The study selection and assessment process were thoroughly illustrated in a PRISMA flow diagram (Figure 1); this refers to the rigorous criteria for inclusion and exclusion. Initially, overall, 541 publications were gathered and evaluated for eligibility. Afterward, the selected studies underwent thorough evaluation using the HWE and NOS scales19, which led to the inclusion of nine studies in this meta-analysis. The characteristics of each study, including their NOS scores, were presented in Table 1. Detailed information on the genotype distribution, allelic frequency, and HWE/Chi-square values of the selected polymorphisms was also provided in Table 2.

None of the five genetic models showed any significant associations. Random effects were implemented due to high heterogeneity, indicated by I² values greater than 50%. In the allelic model, I² was 98%, OR = 1.08, 95% CI [0.35-3.34], p = 0.90; in the homozygous model, I² was 92%, OR = 1.64, 95% CI [0.40-6.71], p = 0.49; in the heterozygous model, I² was 94%, OR = 1.48, 95% CI [0.49-4.42], p = 0.48; in the dominant model, I² was 89%, OR = 1.69, 95% CI [0.61-4.26], p = 0.26; in the recessive model, I² was 97%, OR = 1.00, 95% CI [0.26-3.88], p = 1.00. Forest plots were used to display all the results obtained. A thorough analysis was conducted to determine the sensitivity of interleukin-6 (rs1800797) polymorphisms, which included the execution of Begg's funnel plot and Egger's test. The results showed no publication bias for all five genetic models, as illustrated in Figure 2 and Figure 3.