Prognostic and clinicopathological value of Ki-67 in patients with oesophageal squamous cell carcinoma: a systematic review and meta-analysis


Oesophageal cancer (EC) ranks seventh among different cancers globally and is the sixth most common factor inducing cancer-associated mortality.1 A total of 604 100 newly diagnosed EC cases and 544 076 EC-related deaths were reported globally in 2020.1 Histologically, EC can be classified as adenocarcinoma or oesophageal squamous cell carcinoma (ESCC), and its prevalence varies by geographical region.2 ESCC has the highest prevalence of EC worldwide and accounts for 90% of all EC cases.3 At the time of diagnosis, individuals with ESCC typically exhibit symptoms of severe oesophageal obstruction, and many patients are diagnosed at an advanced stage.4 The mainstay of treatment for locoregional ESCC is surgery. Recent improvements in staging techniques, patient selection, postoperative care and surgical experience have resulted in a marked decrease in surgical morbidity and mortality for ESCC. Neoadjuvant chemotherapy or chemoradiotherapy is the standard treatment for locally advanced ESCC. However, clinical benefits are poor for metastatic or advanced-stage ESCC patients who receive standard treatment, and the median overall survival (OS) is <1 year.5 Consequently, identifying effective biomarkers for predicting ESCC prognosis is essential for improving the clinical outcomes.

As an index related to cell proliferation, Ki-67 expression can be detected in the nucleus, which is tightly associated with cell proliferation and mitosis.6 Ki-67 is related to the cell cycle and mitosis and is commonly used in clinical settings to measure cellular proliferation.7 Ki-67 expression measured through immunohistochemical (IHC) staining has been identified as a biomarker to predict the prognosis of different cancer types, such as breast cancer,8 multiple myeloma,9 cervical cancer,10 diffuse malignant peritoneal mesothelioma11 and oral squamous cell carcinoma.12 Many studies have examined the role of Ki-67 in predicting ESCC prognosis, but no consistent findings have been obtained.13–23 For instance, Ki-67 upregulation has been found to be significantly associated with poor prognosis of ESCC in some studies.18–20 However, other studies have indicated that Ki-67 is not significantly correlated with the prognosis of ESCC.13 16 22 23 Therefore, we collected all eligible studies and performed this meta-analysis to identify the significance of Ki-67 in predicting ESCC prognosis. We also analysed the relationship between Ki-67 expression and the clinicopathological features in the current meta-analysis.

Materials and methods

Patient and public involvement

No patient involved.

Study guideline

The current meta-analysis was carried out in accordance with the guidelines of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (online supplemental file 1).24

Supplemental material

Search strategy

PubMed, Embase, Web of Science and Cochrane Library databases were thoroughly and systematically searched until 26 September 2023 using the search strategies below (Ki67 or Ki-67 or MKI67) and (Esophageal Cancers or Esophageal Neoplasms or Esophageal Cancer or Oesophageal Cancer or Esophagus Cancer or Esophagus Cancers or Cancer of the Esophagus or Esophagus Neoplasms or Cancer of Esophagus or Esophageal Neoplasm or Esophagus Neoplasm or esophageal squamous cell carcinoma or Esophagus or ESCC). The detailed search strategies for each database are shown in online supplemental file 2. Considering that English is the most widely used language in the world and the relatively high quality of English publications in the above-mentioned databases. Only studies published in English were included in this meta-analysis. References and citations in eligible publications were manually searched to identify additional articles.

Supplemental material

Eligibility criteria

The following studies were included: (1) ESCC was diagnosed based on histology or pathology; (2) IHC was performed to measure Ki-67 expression; (3) articles reporting the significance of Ki-67 in predicting one or more survival outcomes, such as OS, progression-free survival, disease-free survival (DFS) or recurrence-free survival (RFS); (4) the HRs and 95% CIs could be available or calculated from the given data; (5) the Ki-67 threshold for stratifying high/low expression was identified and (6) English studies. The following studies were excluded: (1) case reports, letters, meeting abstracts, comments and reviews; (2) articles that included duplicate patients and (3) animal studies.

Data collection and quality evaluation

Two independent researchers (YW and MD were responsible for data extraction from eligible publications. Any disagreement was resolved through negotiations with a third researcher (XC). The following data were collected: first author, publication year, country/region, age, sex, sample size, study design, study period, TNM stage, treatment, Ki-67 threshold, survival outcomes, follow-up, survival analysis type and HRs with 95% CIs. Primary and secondary survival outcomes were OS and DFS, respectively. Study quality was evaluated using the Newcastle-Ottawa Scale (NOS) (range 0–9).25 Articles with NOS scores of ≥6 were considered high-quality studies.

Statistical analysis

The effect of Ki-67 on predicting OS and DFS of ESCC was estimated by calculating pooled HRs and relevant 95% CIs. Between-study heterogeneity was evaluated using Cochrane’s Q test and I2 statistics. Specifically, significant heterogeneities were identified based on p<0.1 on the Q statistic test or I2>50% so the random-effects model should be used; otherwise, the fixed-effects model should be used. Subgroup analyses were conducted according to country/region, sample size, TNM stage, treatment and Ki-67 threshold, and survival analysis was performed to explore the potential sources of heterogeneity. Because this is a meta-analysis study, the per-study sample size could not be properly evaluated. Because the final sample size was determined by the eligible studies included in this meta-analysis, which were selected according to our inclusion and exclusion criteria. In addition, the relationship between Ki-67 and clinicopathological characteristics of ESCC was assessed using pooled ORs and 95% CIs. Furthermore, a sensitivity analysis, in which one study was eliminated each time to examine its impact on pooled results, was conducted to verify the stability of our results. Begg’s and Egger’s tests were used to assess publication bias. Stata V.12.0 software (Stata) was used for the statistical analysis. Differences were considered statistically significant at p<0.05.


Literature retrieval and screening

As shown in figure 1, 2474 studies were identified through primary literature retrieval, and 1791 studies remained after removal of duplicates. Then, 1638 studies were eliminated by screening titles and abstracts because they were irrelevant studies or animal studies, while the full texts of the remaining 153 articles were read. Subsequently, 142 articles were excluded due to a lack of survival data (n=86), not on Ki-67 (n=25), not on ESCC (n=19), overlapping patients recruited (n=6), no Ki-67 threshold (n=3) and no English study (n=3). Ultimately, the present meta-analysis recruited 11 articles including 1124 patients.13–23

Figure 1
Figure 1

The PRISMA flow chart of literature search and study selection. ESCC, oesophageal squamous cell carcinoma; PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analyses.

Enrolled study features

Online supplemental table S1 presents the basic features of the articles.13–23 All studies had a retrospective design and were published between 2005 and 2022. Four studies were performed in Japan,13 14 17 23 four studies were carried out in China16 18 20 22 and one each in Korea,15 Taiwan19 and Thailand,21 respectively. The sample sizes were 25–166 (median 100). The enrolled studies included patients with ESCC. Eight articles included TNM stages I–IV patients,13–20 two enrolled patients with stages II–IV21 23 while one included stages I–III patients.22 IHC was used in all enrolled studies to detect Ki-67 expression. In included studies, the cut-off values of Ki-67 ranged from ≥10% to ≥59%, and the median value was ≥30%. Therefore, we used ≥30% in the following subgroup analysis. Because a cut-off value was identified to dichotomous the patients as high/low Ki-67 expression. Therefore, the percentage is not used as a continuous measure. All 11 articles mentioned the significance of Ki-67 in predicting OS13–23 and 3 studies presented the relationship between Ki-67 and DFS.17 19 23 The HRs and 95% CIs were obtained based on univariate regression in seven articles14 17–19 21–23 and by multivariate analysis in four studies.13 15 16 20 The NOS scores of all enrolled articles were 6–8 (median, 7), indicating high quality (online supplemental table S1).

Supplemental material

Ki-67 and OS in ESCC

All 11 studies comprising 1124 patients13–23 reported the effect of Ki-67 on predicting OS in patients with ESCC. Owing to the obvious heterogeneity (I2=83.7, p<0.001), we applied the random-effects model. As shown in online supplemental table S2 and figure 2, HR 1.62, 95% CI 1.15 to 2.28, p=0.006) were obtained, indicating that increased Ki-67 expression was significantly related to poor OS in ESCC. As shown by subgroup analyses in online supplemental table S2, elevated Ki-67 index still significantly predicted OS in the following subgroups: studies performed in Japan (p=0.001) and Taiwan (p<0.001), sample size≥100 (p=0.039), TNM stages I–IV (p=0.022), surgical treatment (p=0.016), Ki-67 cut-off value >30% (p=0.017) and univariate survival analysis (p<0.001).

Supplemental material

Figure 2
Figure 2

Forest plot of associations between Ki-67 expression and OS of ESCC patients. ESCC, oesophageal squamous cell carcinoma; OS, overall survival.

Ki-67 and DFS in ESCC

Three studies with 216 patients17 19 23 reported the role of Ki-67 in predicting DFS. We used the fixed-effects model because of non-obvious heterogeneity (I2=30.8%, p=0.236). According to our combined data, high Ki-67 expression significantly predicted poor DFS in ESCC (HR 1.72, 95% CI 1.22 to 2.43, p=0.002; online supplemental table S3, figure 3). Based on subgroup analysis, Ki-67 upregulation was significantly related to poor DFS in the following subgroups: studies conducted in Taiwan (p=0.001), sample size≥100 (p=0.001) and Ki-67 cut-off value >30% (p=0.001) (online supplemental table S3).

Supplemental material

Figure 3
Figure 3

Forest plot of associations between Ki-67 expression and DFS of ESCC patients. DFS, disease-free survival; ESCC, oesophageal squamous cell carcinoma.

Correlation of Ki-67 with clinicopathological characteristics of ESCC

Three studies involving 221 patients14 19 23 reported a relationship between Ki-67 expression and clinicopathological features of ESCC. The pooled results demonstrated that Ki-67 was not significantly related to gender (OR 0.74, 95% CI 0.29 to 1.89, p=0.528), tumour differentiation (OR 1.06, 95% CI 0.58 to 1.93, p=0.852), T stage (OR 1.26, 95% CI 0.74 to 2.15, p=0.393), N stage (OR 1.35, 95% CI 0.71 to 2.59, p=0.358), TNM stage (OR 1.61, 95% CI 0.92 to 2.81, p=0.095) and tumour location (OR 0.58, 95% CI 0.24 to 1.42, p=0.235) (figure 4 and online supplemental table S4).

Supplemental material

Figure 4
Figure 4

Forest plots of the relationship between Ki-67 expression and clinicopathological features of ESCC patients. (A) Gender (male vs female); (B) Differentiation (well/moderate vs poor); (C) T stage (T3+T4 vs T1+T2); (D) N stage (N1 vs N0); (E) TNM stage (III+IV vs I+II); and (F) Location (lower+mid-thorax vs upper). ESCC, oesophageal squamous cell carcinoma.

Sensitivity analysis

The stability of the results was assessed using sensitivity analysis (figure 5). Our sensitivity analysis showed that the effect of Ki-67 in predicting OS and DFS was not significantly influenced by any single study (figure 5).

Figure 5
Figure 5

Sensitivity analysis. (A) OS and (B) DFS. DFS, disease-free survival; OS, overall survival.

Publication bias

We used Begg’s and Egger’s tests to evaluate possible publication bias. Online supplemental figure S1 reveals symmetry of funnel plots, indicating the absence of obvious publication bias for OS (p=0.876 and 0.917 on Begg’s and Egger’s tests separately) and DFS (p=0.602 and 0.675 on Begg’s and Egger’s tests separately).

Supplemental material


The significance of Ki-67 in predicting ESCC remains controversial according to previous studies. In this meta-analysis, we synthesised information from 11 articles involving 1124 patients in this meta-analysis.13–23 As a result, the increased Ki-67 index significantly predicted the OS and DFS of patients with ESCC. Subgroup analysis indicated that Ki-67 upregulation significantly predicted OS and DFS when using a Ki-67 threshold of >30%. Nonetheless, Ki-67 expression was not significantly correlated with the clinicopathological characteristics of ESCC, possibly due to the limited sample size. Taken together, this meta-analysis revealed the role of Ki-67 upregulation in predicting short-term and long-term survival of patients with ESCC. To our knowledge, this meta-analysis is the first to investigate the prognostic and clinicopathological significance of Ki-67 expression in ESCC patients.

K-i67 was first discovered by Gerdes et al.26 As a method for detecting cell proliferation indirectly, Ki-67 indicates cancer cell proliferation ability, and it can only be detected within proliferating cells.27 According to our results, Ki-67 is significant for predicting ESCC prognosis, and the potential underlying mechanisms are as follows. First, cell proliferation can be evaluated based on Ki-67 expression in every active cell cycle phase, such as G1, S, G2 or mitosis.28 Thus, increased Ki-67 expression promotes tumour growth and increases tumour volume, which is more likely to lead to worse prognosis. Second, epithelial-to-mesenchymal transition (EMT) may also be influenced by Ki-67, apart from the frequently used proliferation marker.6 29 EMT induction is crucial for the occurrence and development of cancers. Therefore, it is rational to use Ki-67 as a prognostic biomarker for ESCC, based on its biological function.

Notably, all eligible studies are from Asian regions in this meta-analysis. Four studies were performed in Japan,13 14 17 23 four in China,16 18 20 22 and one each in Korea,15 Taiwan19 and Thailand,21 respectively. Actually, we did not restrict the geographical regions of eligible studies and we searched English literature in four databases. This issue can be explained in the following aspect. First, EC incidence varies widely by geography globally, with a 60-fold difference between regions with high and low incidence.30 The age-standardised incidence rate of EC was highest in Eastern Asia (12.2 per 100 000 people).31 From northern Iran and Central Asia to northern China, the region with the highest incidence of EC is often referred to as the ‘EC belt’.32 33 Second, in the world, approximately 87% of EC cases are ESCC, while 11% are oesophageal adenocarcinoma (EAC).34 ESCC is the most prevalent histological type worldwide while EAC is predominant in the USA and other Western countries.35 36 Because we included patients with ESCC in this meta-analysis, the cases of EAC are ineligible. Therefore, the cases from Western countries could be small because EAC is predominant in these regions.

Recently, Ki-67 has been demonstrated to have a significant prognostic impact on various cancers through a meta-analysis.37–39 Moltajaei et al showed that Ki-67 upregulation might predict dismal prognostic outcomes in patients with oral melanoma in a meta-analysis including six studies.37 According to Li et al, Ki-67 upregulation may be a useful biomarker of the risk of recurrent gastrointestinal stromal tumour transformation in a meta-analysis involving 1682 patients.39 Another recent meta-analysis comprising 13 studies revealed that Ki-67 upregulation was related to poor OS, DFS and local RFS in nasopharyngeal carcinoma.40 In a meta-analysis involving 38 articles conducted by Qiu et al, Ki-67 upregulation was markedly associated with poor OS, which is a potential prognostic marker in ovarian cancer.41 Moreover, in a meta-analysis involving 5398 patients, Ki-67 upregulation was related to poor OS, DFS and cancer-specific survival in renal cell carcinoma.42 Therefore, the findings of the current meta-analysis confirmed the impact of Ki-67 on predicting the prognosis of additional solid tumours.

In this meta-analysis, the HRs for OS and DFS were 1.62 and 1.72, respectively. The estimated effect on OS and DFS was modest and the prognostic power of Ki-67 needs to be improved. Notably, there are many other prognostic biomarkers for EC according to previous studies.43 There are non-invasive biomarkers including blood, plasma, saliva, and urine biomarkers,43 molecular biomarkers, and imaging-based biomarkers. Ki-67 is a part of these prognostic markers for EC cases. Additionally, the prognostic efficiency of Ki-67 can be enhanced in combination with other non-invasive biomarkers such as circulating tumour cells, circulating tumour DNA and neutrophil-to-lymphocyte ratio.44 45

Some limitations of this study should be noted. First, our included articles had a retrospective design, which probably introduced a selection bias. Because of the inherent nature of retrospective studies, selection bias may exist. Second, the Ki-67 threshold was not consistent among the enrolled studies. Therefore, the criteria for identifying high/low Ki-67 expression may vary across eligible studies. Third, all the included studies were from Asian regions, which may be due to the high incidence of ESCC in Asia. Therefore, our results may be applicable to Asian ESCC cases. Because of these limitations, more prospective multiregional studies should be conducted for further validation.

This post was originally published on