Page 111 - GPD-4-1

Gene & Protein in Disease Prognostic role of SIRT1 expression in cancer

2.3. Data extraction heterogeneity were also conducted. Bubble plots were

The included studies were independently reviewed by three generated to visually represent the relationship between
authors – MI, ST, and SA – for data extraction. Various moderators and effect sizes. Publication bias was assessed
variables were extracted using Google Forms, which were through Egger’s regression-based test, including intercept
later exported into an Excel spreadsheet in MS Office 2021 regression and statistical estimation based on t-statistics,
v2310 (Microsoft, United States). The extracted variables examining the effect size and standard errors of individual
included: (i) Study characteristics (author, year, country, studies. Finally, bubble and funnel plots were drawn to
sample size, cancer type, etc.), (ii) patient characteristics further assess the results.
(age, sex, tumor stage, etc.), (iii) SIRT1 expression levels, 3. Results
and (iv) OS rates. Standardized categories were used for
tumor stage and sample size to ensure consistency across Upon evaluation, 15 studies (total number of cases = 3059)
studies. Studies with incomplete data, such as missing key out of 158 eligible studies met the inclusion criteria
prognostic details (e.g., tumor stage or survival rates), were (Figure 1) and were selected for analysis (Table 1). The
excluded from the final analysis. The values for all variables majority of studies were conducted in Asia (n = 10),
were then standardized and formatted for further analysis specifically China (n = 5), while only five studies were
by the authors AA and MBK. conducted in non-Asian regions. The included studies
covered seven cancer types: ovarian (n = 3), uterine (n = 2),
2.4. Quality appraisal of studies renal cell (n = 2), lung (n = 3), gastric (n = 2), breast (n = 1),
Two authors, UAKS and FI, independently assessed the and colorectal cancer (n = 2). Most studies employed
quality of the included studies following the guidelines immunohistochemistry to analyze SIRT1 expression, with
35
of the Joanna Briggs Institute. The evaluation involved the exception of Tan et al., who used gene set enrichment
checking the relevance of each study against specific criteria analysis. The majority of studies (n = 8) had smaller sample
(Supplementary File for detailed quality appraisal criteria). sizes (<100), while five studies had medium sample sizes
The focus was on studies examining how SIRT1 influences (≈100 – 500), and only two studies had larger sample sizes
the prognosis of various human cancers. This thorough (>500).
review process included comprehensive assessments of 3.1. Overall findings
the following: sample frame appropriateness, participant
sampling, adequacy of sample size, detailed descriptions The current meta-analysis, using a random-effects
of subjects and settings, coverage of data analysis, validity model to account for potential variability between
of methods for condition identification, standardization studies, revealed a significant prognostic effect for SIRT1
and reliability of condition measurement, appropriateness (pooled HR = 1.483 ± 0.2974, 95% confidence interval (CI):
of statistical analysis, and adequacy of response rate. Any 0.900 – 2.065), as depicted in Figure 2. This pooled effect size
disagreements were resolved through discussion, and the demonstrated a strong, statistically significant, and positive
corresponding authors, AA and MBK, were consulted. prognostic value for SIRT1. The statistical significance was
underscored by a Z-score of 4.98 (P < 0.001), indicating
2.5. Statistical analysis that SIRT1 has a substantial impact in the context of the
Meta-analysis was conducted using SPSS Statistics v28.0.1.1 analyzed studies. However, the heterogeneity statistics
2
2
2
(IBM, United States) under a subscription-based license, (I = 0.00, specifically, H = 1.00 and Tau = 0.0001) in our
with pre-calculated effect sizes, primarily hazard ratios meta-analysis indicated a remarkably low level of variability
(HRs) for OS rates, and the selection of a random effects among the included studies. In addition, the homogeneity
2
model. A forest plot was generated for the overall analysis. analysis showed high homogeneity (I = 0.0001), further
Subgroup analyses, based on each variable, were performed supporting the robustness and consistency of the findings.
using the same protocol. Meta-regression was conducted These results underscore the significance of SIRT1,
to investigate sources of heterogeneity, with variables such demonstrating its substantial impact as a prognostic factor
as study location, sample size categories, cancer types, in the analyzed studies.
age groups, tumor stage, and SIRT1 expression included
as moderators. The criteria for assessing heterogeneity 3.2. Confirmatory analysis for heterogeneity
included Cochran’s Q-statistics and the I² statistic. Small- The presence of heterogeneity in the meta-analysis was
study effects were evaluated using Egger’s regression-based extensively examined for various potential moderators, as
test, where the intercept of the regression line was used to shown in Table 2, including country (Figure 3A), sample
assess asymmetry in the funnel plot, which could indicate size categories (Figure 3B), cancer type (Figure 3C), age
potential bias. Tests of residual homogeneity and residual group (Figure 3D), tumor stage (Figure 3E), and SIRT1

Volume 4 Issue 1 (2025) 3 doi: 10.36922/gpd.4294