Prognostic and clinicopathological effect of the prognostic nutritional index (PNI) in patients with cervical cancer: a meta-analysis

Abstract

Background

Numerous studies have explored whether the prognostic nutritional index (PNI) can predict the prognosis of cervical cancer (CC); however, their findings remain controversial. This meta-analysis focused on evaluating the relationship between the PNI and the prognosis of patients with CC.

Methods

Relevant articles were collected from specific databases up to March 16, 2023. The relationship between the PNI and survival outcomes in patients with CC was estimated using combined hazard ratios (HRs) and associated 95% confidence intervals (CIs). The association of the PNI with clinicopathological features in patients with CC was assessed by combining odds ratios (ORs) and associated 95% CIs.

Results

Nine articles with 2508 cases were included in the meta-analysis. According to our pooled findings, a decreased PNI showed a significant association with worse overall survival (OS) (HR = 2.98, 95% CI = 2.22–3.99, p < .001) as well as progression-free survival (PFS) (HR = 2.43, 95% CI = 1.92–3.07, p < .001) in patients with CC. The subgroup analysis indicated that the results were reliable. Moreover, the decreased PNI showed a significant association with the presence of lymph node metastasis (LN metastasis, OR = 1.53, 95% CI = 1.04–82.24, p = .030) and maximum tumor size >4 cm (OR = 1.73, 95% CI = 1.21–2.46, p = .002). However, the PNI was not significantly associated with histology, differentiation, or FIGO stage.

Conclusion

In this study, a low PNI predicted dismal OS and PFS in patients with CC, who also tend to suffer from LN metastasis and larger tumor size. PNI is a promising biomarker for predicting the prognosis of patients with CC in clinical practice.

KEY MESSAGES

1. To our knowledge, the present meta-analysis is the first to explore whether the PNI can be used to predict the prognosis of patients with CC.

2. In this study, a low PNI predicted dismal OS and PFS in patients with CC, who also tend to suffer from LN metastasis and larger tumor size.

3. PNI is a promising biomarker for predicting the prognosis of patients with CC in clinical practice.

Keywords:

• Cervical cancer
• meta-analysis
• prognostic nutritional index
• prognosis
• clinical practice

Introduction

Cervical cancer (CC) is a major global threat to the health of women [1] and ranks fourth among cancers affecting women worldwide [2]. According to the GLOBOCAN estimates, in 2020, 604,127 newly diagnosed cases and 341,831 deaths due to CC occurred worldwide [3]. Despite advances in the treatment of CC, its prognosis remained poor over the past several decades, especially in developing countries. CC has a poor prognosis primarily due to poor screening and treatment of early-stage disease, although it is one of the most curable cancers with a natural history quite well-known. Consequently, the identification of new and effective biomarkers to predict the prognosis of patients with CC is urgently needed.

Accumulating evidence has shown a significant connection between nutritional status, operative complications, and clinical outcomes in patients with cancer [4]. Many inflammatory blood-based parameters such as systemic immune-inflammation index (SII) [5], controlling nutritional status (CONUT) score [6], geriatric nutritional risk index (GNRI) [7], and albumin-to-globulin ratio (AGR) [8] have been reported as effective prognostic markers in patients with cancer. The prognostic nutritional index (PNI) can be computed based on serum albumin levels and peripheral blood lymphocyte counts [9]. The PNI was first proposed by Buzby et al. in 1980 to evaluate the risk of postoperative complications in patients undergoing gastrointestinal surgery [9]. PNI was calculated using the following formula: PNI = peripheral blood (10 × serum albumin [g/dL] + 0.005 × total lymphocyte count [/mm³]) [10]. Recently, numerous studies have indicated that PNI can be used to predict the prognosis of patients with different types of cancer, including gastric cancer [11], hepatocellular carcinoma [12], lymphoma [13], urothelial carcinoma [14], and pancreatic neuroendocrine neoplasms [15]. Specifically, there is evidence that the lower the PNI before treatment, the worse the prognosis of patients with prostate cancer treated with androgen deprivation therapy [16]. The pretreatment-PNI is an independent prognostic factor, irrespective of neoadjuvant chemotherapy, for patients with esophageal squamous cell carcinoma [17]. Current evidence also suggests that the PNI is a significant prognostic marker for disease-specific and event-free survival in patients with soft tissue sarcomas [18]. Additionally, a multivariate model showed that a low PNI was associated with shorter survival duration in patients with follicular lymphoma [19]. Although the effect of PNI on predicting the prognosis of patients with CC has been widely explored [20–28], the findings remain conflicting. Some studies have reported that a decreased PNI predicts worse survival in patients with CC [20,23,24], whereas others have indicated no obvious correlation between PNI and CC prognosis [21]. Therefore, we collected updated data for the present meta-analysis to accurately identify the role of PNI in predicting the prognosis of patients with CC.

Materials and methods

Study guideline

This meta-analysis was performed following the guidelines of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) [29].

Search strategy

The Embase, PubMed, Cochrane Library, Web of Science, and China National Knowledge Infrastructure databases were systematically searched. The search terms were as follows: (prognostic nutritional index OR PNI) AND (cervical carcinoma OR cervical neoplasm OR cervical cancer OR cervical tumor). Literature retrieval was updated on March 16, 2023. There were no restrictions on publication language. Additionally, we manually checked the reference lists and related reviews to identify potential inclusions.

Inclusion and exclusion criteria

Patients included should satisfy the following criteria: (1) CC was pathologically diagnosed; (2) articles reporting the relationship of PNI with survival outcomes in patients with CC; (3) cut-off values should be identified to divide low/high PNI; (4) hazard ratios (HRs) and associated 95% confidence intervals (CIs) of survival outcomes should be available or calculable; and (5) available survival outcomes, such as overall survival (OS), progression-free survival (PFS), cancer-specific survival (CSS), and recurrence-free survival (RFS). Studies conforming to the following criteria were excluded: (1) reviews, letters, meeting abstracts, comments, and case reports; (2) duplicates; and (3) animal studies.

Data extraction and quality assessment

Two reviewers (Z.N. and B.Y.) independently assessed and collected information from each included study. Any disagreement was resolved through discussions to reach a consensus. The following information was collected in each study: first author; publication year; country; sample size; age; study period; International Federation of Gynecology and Obstetrics (FIGO) stage; treatment; follow-up; threshold; threshold calculation approach; survival endpoints; survival analysis type; study design; HRs; and 95% CIs. The quality of the studies was evaluated using the Newcastle-Ottawa Quality Assessment Scale (NOS), which includes three aspects: selection, outcomes, and comparability, with a total score of 0–9. Studies with NOS scores ≥6 were considered to be of high quality.

Statistical analysis

This study calculated pooled HRs and 95% CIs to estimate the relationship between the PNI and the survival outcomes of patients with CC. Heterogeneities among articles were assessed by I² statistics and Cochran’s Q test, with P heterogeneity <0.10 and I² > 50% indicating heterogeneity. Therefore, the random-effects model was applied to this condition; otherwise, the fixed-effects model was adopted. Possible sources of heterogeneity were detected using subgroup analysis. The association of the PNI with clinicopathological characteristics in patients with CC was evaluated through combining odds ratios (ORs) and 95% CIs. Begg’s and Egger’s tests were used to detect possible publication biases. Stata software (version 12.0; Stata Corp., College Station, TX, USA) with a copyright license obtained was used for the statistical analysis. Statistical significance was set at p < .05 (two-sided).

Results

Process of literature search

A total of 265 records were detected through the primary search, of which duplicate articles were eliminated, and 163 articles remained (Figure 1). After reading their titles and abstracts, 146 studies were eliminated because of irrelevance, and 17 studies were further evaluated. The full texts of these 17 articles were read, and eight were eliminated for the following reasons: no survival data provided (n = 5), overlapped patients (n = 2), and no PNI analysis (n = 1). Ultimately, this study included nine articles involving 2508 cases [20–28] (Figure 1).

Figure 1. PRISMA Flow chart of the data search.

Features of the included articles

Table 1 shows the basic features of the selected articles [20–28] published from 2016 to 2023. Eight studies were performed in China [21–28] and one in Japan [20]. All the studies were retrospective in design. Six studies were published in English [20–22,24,25,28] and three studies were published in Chinese [23,26,27]. The median sample size included in these articles was 229 patients (range, 82–698). Four studies included patients with FIGO stages I–II [22,24,25,28], four studies included those with stages I–IV [20,21,23,27], and one study included those with stages I–III [26]. The median PNI cut-off value was 48.82 (range, 45–52.68). Seven articles [20,22–24,26–28] selected thresholds via receiver operating characteristic (ROC) curves, and two studies [21,25] determined the cut-off values based on the literature. Eight studies reported that the PNI could be used to predict OS [20–26,28], whereas six studies [20,22–25,27] reported a relationship between the PNI and PFS in patients with CC. The median NOS score was 7 (range, 6–9), suggesting the high quality of each included article.

Table 1. Basic characteristics of included studies in this meta-analysis.

Study	Year	Country	Sample size	Age (years) Median(range)	Study duration	FIGO stage	Treatment	Follow-up (months) Median(range)	Cut-off value	Cut-off determination	Survival endpoint	Survival analysis	NOS score
Haraga, J.	2016	Japan	131	61.5(25–88)	2007–2013	I–IV	CCRT	60	48.55	ROC curve	OS, PFS	Multivariate	7
He, X.	2018	China	229	44(28–79)	2007–2009	I–IV	Surgery+ CCRT	83	45	Literature	OS	Univariate	7
Zhang, W.	2018	China	235	46(29–78)	2005–2009	I–II	Mixed	77(32–96)	50.38	ROC curve	OS, PFS	Univariate	8
Ren, J. L.	2020	China	324	51.6(25–84)	2015–2016	I–IV	CCRT	39(4–59)	49.55	ROC curve	OS, PFS	Multivariate	7
Jiang, Y. A.	2021	China	583	≤45 y: 203 >45 y: 380	2009–2017	I–II	Surgery	68	50.15	ROC curve	OS, PFS	Univariate	9
Zhang, G.	2021	China	698	51	2004–2015	I–II	Surgery+ CCRT	56.2(4.9–186.9)	48.55	Literature	OS, PFS	Multivariate	8
Zhang, Z. C.	2022	China	82	50.13(40–60)	2015–2017	I–III	Surgery+ CCRT	1–36	48.82	ROC curve	OS	Univariate	6
Zhu, J. M.	2022	China	117	<60 y: 70 ≥60 y: 47	2019–2021	I–IV	Surgery+ CCRT	1–12	46.2	ROC curve	PFS	Univariate	7
Guo, J.	2023	China	109	53.95(27–71)	2014–2017	I–II	Surgery	1–96	52.68	ROC curve	OS	Univariate	7

FIGO: International Federation of Gynecology and Obstetrics; CCRT: concurrent chemoradiotherapy; ROC: receiver operating characteristics; OS: overall survival; PFS: progression-free survival; NOS: Newcastle-Ottawa Scale.

PNI and OS of patients with CC

Eight studies involving 2,391 patients [20–26,28] reported a relationship between the PNI and OS in patients with CC. A fixed-effects model was used because of insignificant heterogeneity (I² = 6.1%, p = .383). According to Figure 2 and Table 2, pooled HR = 2.98, 95% CI = 2.22–3.99, and p < .001, which indicated a significant relation between decreased PNI and worse OS among patients with CC. Subgroup analysis stratified by various factors was conducted for OS. As shown in Table 2, a decreased PNI significantly predicted OS regardless of country, sample size, FIGO stage, threshold, or survival analysis type (p < .05). Furthermore, decreased PNI markedly predicted dismal OS when cut-off value was determined by ROC analysis (HR = 3.69, 95% CI = 2.63–5.18, p < .001; Table 2).

Figure 2. Forest plots of prognostic role of PNI for OS in patients with cervical cancer.

Table 2. Subgroup analysis of prognostic role of PNI for OS in patients with cervical cancer.

	No. of	No. of				Heterogeneity
Subgroup	studies	patients	HR (95%CI)	p	Effects model	I^2(%)	Ph
Total	8	2391	2.98(2.22–3.99)	<.001	Fixed	6.1	0.383
Country
China	7	2260	2.93(2.14–4.02)	<.001	Fixed	18.8	0.286
Japan	1	131	3.27(1.48–7.24)	.003	–	–	–
Sample size
≤230	4	551	3.33(1.93–5.75)	<.001	Fixed	0	0.869
>230	4	1840	2.81(1.65–4.79)	<.001	Random	53.9	0.089
FIGO stage
I–II	4	1625	2.39(1.56–3.67)	<.001	Fixed	37.4	0.188
I–III	1	82	3.21(1.02–10.07)	.046	–	–	–
I–IV	3	684	3.68(2.39–5.67)	<.001	Fixed	0	0.699
Cut-off value
≤49	4	1140	2.16(1.39–3.35)	.001	Fixed	4.4	0.371
>49	4	1251	3.87(2.61–5.75)	<.001	Fixed	0	0.907
Cut-off determination
ROC curve	6	1464	3.69(2.63–5.18)	<.001	Fixed	0	0.980
Literature	2	927	1.54(0.85–2.78)	.152	Fixed	0	0.541
Survival analysis
Multivariate	3	1153	2.70(1.37–5.34)	.004	Random	68.8	0.041
Univariate	5	1238	3.33(2.06–5.37)	<.001	Fixed	0	0.949
Treatment
Surgery + CCRT	3	1009	1.80(1.06–3.04)	.029	Fixed	0	0.446
CCRT	2	455	3.86(2.46–6.07)	<.001	Fixed	0	0.618
Surgery	2	692	3.72(1.88–7.33)	<.001	Fixed	0	0.555
Mixed	1	235	3.24(1.16–9.06)	.025	–	–	–

FIGO: International Federation of Gynecology and Obstetrics; ROC: receiver operating characteristics; OS: overall survival.

PNI and PFS in patients with CC

Six studies consisting of 2088 patients [20,22–25,27] provided data on whether the PNI could be used to predict PFS among patients with CC. A pooled HR = 2.43, 95% CI = 1.92–3.07, and p < .001 were obtained, indicating that a low PNI predicted the shortened PFS among patients with CC (Table 3; Figure 3). As revealed by subgroup analysis, a reduced PNI significantly predicted PFS in all subgroups, including sample size, country, FIGO stage, threshold, threshold calculation, and survival analysis type (all p < .05; Table 3).

Figure 3. Forest plots of prognostic role of PNI for PFS in patients with cervical cancer.

Table 3. Subgroup analysis of prognostic role of PNI for PFS in patients with cervical cancer.

	No. of	No. of				Heterogeneity
Subgroup	studies	patients	HR (95%CI)	p	Effects model	I^2(%)	Ph
Total	6	2088	2.43(1.92–3.07)	<.001	Fixed	35.4	0.171
Country
China	5	1957	2.43(1.90–3.12)	<.001	Fixed	48.3	0.102
Japan	1	131	2.38(1.21–4.67)	.012	–	–	–
Sample size
≤230	2	248	2.97(1.69–5.23)	<.001	Fixed	27.6	0.240
>230	4	1,840	2.33(1.80–3.01)	<.001	Fixed	48.0	0.123
FIGO stage
I–II	3	1516	1.85(1.32–2.58)	<.001	Fixed	0	0.510
I–IV	3	572	3.14(2.27–4.36)	<.001	Fixed	0	0.486
Cut-off value
≤49	3	946	2.31(1.31–4.06)	.004	Random	52.3	0.123
>49	3	1142	2.84(2.06–3.90)	<.001	Fixed	0	0.460
Cut-off determination
ROC curve	5	1390	2.87(2.17–3.79)	<.001	Fixed	0	0.566
Literature	1	698	1.61(1.04–2.49)	.032	–	–	–
Survival analysis
Multivariate	3	1153	2.32(1.46–3.68)	<.001	Random	62.5	0.069
Univariate	3	935	2.66(1.66–4.26)	<.001	Fixed	9.5	0.331
Treatment
Surgery + CCRT	2	815	2.56(0.86–7.64)	.091	Random	74.3	0.049
CCRT	2	455	2.99(2.12–4.22)	<.001	Fixed	0	0.443
Surgery	1	583	2.00(1.05–3.81)	.035	–	–	–
Mixed	1	235	2.90(1.15–7.32)	.024	–	–	–

FIGO: International Federation of Gynecology and Obstetrics; ROC: receiver operating characteristics; PFS: progression-free survival.

Relation of PNI with clinicopathological features

This study investigated the relationship between the PNI and clinicopathological factors of CC using data from four studies involving 1,098 patients [23,24,26,28]. The following factors were analyzed: histology [non-SCC (squamous cell carcinoma) vs SCC], lymph node (LN) metastasis (yes vs no), differentiation (poor vs well/moderate), maximal tumor size (>4 cm vs ≤4 cm), and FIGO stages II vs I and III–IV vs I–II. According to Table 4 and Figure 4, the combined results indicated the relation of PNI with the presence of LN metastasis (OR = 1.53, 95% CI = 1.04–82.24, p = .030) and maximum tumor size >4 cm (OR = 1.73, 95% CI = 1.21–2.46, p = .002). However, PNI was not significantly related to histology (OR = 1.41, 95% CI = 0.66–2.98, p = .374), differentiation (OR = 0.76, 95% CI = 0.51–1.12, p = .168), and FIGO stage (I–II) (OR = 0.90, 95% CI = 0.61–1.35, p = .624) or (I–IV) (OR = 1.97, 95% CI = 0.70–5.565, p = .199) (Figure 4; Table 4).

Figure 4. Forest plots of association between PNI and clinicopathological factors in patients with cervical cancer. (A) histology; (B) lymph node metastasis; (C) differentiation; (D) maximum tumor size; (E) FIGO stage (II vs I); and (F) FIGO stage (III–IV vs I–II).

Table 4. The correlation between PNI and clinicopathological features in patients with cervical cancer.

	No. of	No. of				Heterogeneity
Variables	studies	patients	OR (95%CI)	p	Effects model	I^2(%)	Ph
Histology (non-SCC vs SCC)	3	989	1.41(0.66–2.98)	.374	Random	52.7	0.120
LN metastasis (yes vs no)	3	515	1.53(1.04–82.24)	.030	Fixed	0	0.382
Differentiation (poor vs well/moderate)	2	665	0.76(0.51–1.12)	.168	Fixed	0	0.982
Maximum tumor size (cm) (>4 vs ≤4)	2	907	1.73(1.21–2.46)	.002	Fixed	0	0.648
FIGO stage (II vs I)	2	692	0.90(0.61–1.35)	.624	Fixed	0	0.478
FIGO stage (III–IV vs I–II)	2	406	1.97(0.70–5.56)	.199	Random	73.7	0.051

SCC: squamous cell carcinoma; LN: lymph node.

Publication bias

Begg’s and Egger’s tests were used to assess the possible publication bias. Funnel plots did not exhibit any symmetry (Figure 5), indicating the absence of an obvious publication bias in this meta-analysis: p = .536 and 0.819 upon Begg’s and Egger’s tests separately for OS, p = .452 and .573 for PFS separately.

Figure 5. Publication bias test by using Begg’s test and Egger’s test. (A) Begg’s test for OS, p = .536; (B) Egger’s test for OS, p = .819; (C) Begg’s test for PFS, p = .452; (D) Egger’s test for PFS, p = .573.

Discussion

To our knowledge, the present meta-analysis is the first to explore whether the PNI can be used to predict the prognosis of patients with CC. The PNI reflects the nutritional and immune statuses of individual patients. Whether PNI can be used to predict the prognosis of CC remains controversial, according to previous investigations. In this meta-analysis, we collected data from nine studies with 2508 patients to accurately explore the effect of PNI on predicting the prognosis of patients with CC. According to our meta-analysis, a decreased PNI predicted poor OS and PFS among patients with CC. The PNI reliably predicted the prognosis of patients with CC in various subgroups. Moreover, our study also indicated that decreased PNI was associated with LN metastasis and tumor size >4 cm, suggesting that patients with CC with low PNI are prone to tumor metastasis and have a higher tumor burden. Taken together, the current meta-analysis revealed that the PNI is a cost-effective and credible prognostic marker for patients with CC.

Nutrition and inflammation play pivotal roles in tumorigenesis, cancer progression, and metastasis by modulating the tumor microenvironment [30–32]. PNI was calculated using serum albumin levels and lymphocyte counts; therefore, a low PNI can be attributed to a low serum albumin level and lymphocyte count. The potential mechanisms of the prognostic role of PNI in patients with CC are interpreted below. Most serum proteins are composed of albumin, which is produced in the liver. Albumin has a critical effect on human body functions [8]. Serum albumin is an acute-phase protein related to systemic inflammation that reflects nutritional status [33]. The level of serum albumin is related to chronic inflammation, which activates cytokines such as IL-1 and TNF-α [34–37]. Second, the blood lymphocyte levels were calculated as part of the PNI. As a fundamental component of cell-mediated immunity, lymphocytes inhibit the proliferation and invasion of tumor cells via cytokine-mediated cytotoxicity [38,39]. Several lymphocyte subpopulations have also been implicated in the regulation of tumor progression, including B cells, natural killer (NK) cells, and CD4+ and CD8+ T cells [40]. Moreover, malnutrition can further weaken the immune system, increase side effects, and render multiple treatment strategies less effective [41,42]. Therefore, the PNI combines albumin level and lymphocyte count and is a powerful indicator of the prognosis of patients with CC.

Notably, the results indicated that a low PNI was significantly associated with the presence of LN metastasis and tumor size of > 4 cm (Table 4). Disease severity is associated with a lower PNI in more advanced tumors. It has long been considered that albumin does not reflect the nutritional status of patients with cancer but rather cancer severity. Therefore, the PNI cannot be considered a tool that can detect advanced cancer stages, but rather a prognostic tool that can correlate cancer severity with a poorer prognosis. However, a worse prognosis is expected in patients with more advanced tumors. Therefore, it is important to investigate the prognostic role of the PNI in survival stratified by tumor staging. The results of the subgroup analysis in our meta-analysis showed that the PNI remained a significant prognostic factor for OS and PFS, irrespective of tumor stage (Tables 2 and 3).

Recently, many meta-analyses have reported that the PNI can be used to predict the prognosis of various solid tumors [43–48]. In a recent meta-analysis of 2, 322 cases of breast cancer, low pretreatment PNI deteriorated OS as well as disease-free survival (DFS) rates [44]. Kang et al. have reported that patients with gastrointestinal stromal tumors who had a decreased PNI showed poor OS and RFS in a meta-analysis of 2307 patients [46]. In another meta-analysis that recruited 3, 631 patients, Tang et al. have showed that patients with nasopharyngeal carcinoma with decreased PNI had poor OS, PFS, and distant metastasis-free survival [48]. In a study conducted by Luan et al. a decreased PNI predicted dismal OS and worse PFS in patients with diffuse large B-cell lymphoma [49].

This study had certain limitations. First, the enrolled articles were from East Asian countries, particularly China. Therefore, our findings apply to Asian patients with CC. Second, the PNI thresholds were not uniform among the included studies. Third, all included studies were retrospective, which may have caused a selection bias. Therefore, large multicenter clinical studies should be conducted for validation.

Conclusions

In conclusion, the PNI significantly predicted poor OS and PFS in patients with CC. Patients with CC and a decreased PNI also tend to suffer from the presence of LN metastasis and larger tumor size. The PNI is a promising prognostic marker for patients with CC in clinical practice.

Authors contributions

ZN contributed the central idea, analyzed most of the data and wrote the main manuscript text. ZN and BY contributed to refining the ideas, collecting the data, carrying out additional analyses and revising the manuscript. All authors contributed to the article and approved the submitted version.

Ethics statement

Ethical approval was not required for this meta-analysis because all the data used in this study were obtained from published articles.

Acknowledgements

We would like to thank Editage (www.editage.com) for English language editing.

Disclosure statement

The authors declare that there is no conflict of interest.

Data availability statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Additional information

Funding

This work was supported by grants from the National Natural Science Foundation of China (82172361, 82372329), Zhejiang Province Key Research and Development Project (2023C03048), and Medicine and Health Project of Zhejiang Province (2018KY276).