J Cancer Res Clin Oncol DOI 10.1007/s00432-014-1714-3
Review - Clinical Oncology
The prognostic significance of the prognostic nutritional index in cancer: a systematic review and meta‑analysis Kaiyu Sun · Shuling Chen · Jianbo Xu · Guanghua Li · Yulong He
Received: 19 April 2014 / Accepted: 13 May 2014 © Springer-Verlag Berlin Heidelberg 2014
Abstract Purpose The prognostic nutritional index (PNI) is a simple and effective parameter, initially created to evaluate preoperative nutritional conditions and surgical risk. It has been recently been found to be associated with shortand long-term outcomes of various malignancies. We performed a meta-analysis to determine the predictive significance of PNI in cancer, as a mean to assist in determining the optimal surgery timing and in improving the survival of cancer patients. Methods Data were retrieved from PubMed and ISI Web of Science to identify eligible studies. Odds ratios (ORs) and hazard ratios (HRs) were extracted and pooled to explore the relationships of PNI with patient survival and clinicopathological features. Results Fourteen studies with a total of 3,414 participants met the inclusion criteria. Low PNI was associated with poor overall survival (pooled OR 1.80, 95 % confidence interval [CI] 1.59–2.04) and the presence of post-operative complications (pooled OR 2.45, 95 % CI 1.31–4.58) in cancer patients, but not with cancer-specific survival (CSS) (pooled HR 1.81, Kaiyu Sun and Shuling Chen have contributed equally to the work. K. Sun · J. Xu · G. Li · Y. He (*) Division of Gastrointestinal Surgery, The First Affiliated Hospital, Sun Yat-sen University, Guangzhou 510080, People’s Republic of China e-mail: [email protected] K. Sun e-mail: [email protected] S. Chen Division of Gastroenterology, The First Affiliated Hospital, Sun Yat-sen University, Guangzhou 510080, People’s Republic of China
95 % CI 0.94–3.49). PNI was also found to be associated with invasion depth (pooled OR 5.07, 95 % CI 2.34–10.96) and lymph node metastasis (pooled OR 3.70, 95 % CI 2.32– 5.92) in gastric cancer, whereas TNM stage was the only clinicopathological feature associated with PNI in colorectal carcinoma (pooled OR 1.81, 95 % CI 1.24–2.64). Conclusions PNI might be an effective predictive indicator for the prognosis of cancer, especially digestive system carcinomas. Further studies are required to verify the significance of PNI in clinical practice. Keywords Prognostic nutritional index · Cancer · Prognosis · Meta-analysis Abbreviations CIs Confidence intervals CSS Cancer-specific survival HR Hazard ratio HRs Hazard ratios OR Odds ratio ORs Odds ratios OS Overall survival PNI Prognostic nutritional index
Introduction Cancer has always represented a major medical problem and is associated with high morbidity, mortality, and economic burden, despite great improvements in early detection, surgical techniques, chemotherapy, radiotherapy, biological treatment, and multidisciplinary treatment in recent years (Jemal et al. 2011). Surgery remains the most important tool for the treatment of solid tumours, if allowed by the condition of the patients. However, many patients develop
13
post-operative complications or even relapse after surgery. The preoperative conditions of the patients, especially the nutritional and immunological conditions, have been found to be associated with both the post-operative prognosis and the long-term outcomes of malignant tumours (Schwegler et al. 2010). Therefore, it is crucial to identify an effective nutritional and immunological index to predict patient prognosis in order to establish the optimal preoperative medical treatment and to determine the optimal time to operate. Several nutritional and immunological indices to predict cancer patients’ prognoses have been identified (Nozoe et al. 2010; Asher et al. 2011; Chua et al. 2011; Li et al. 2014). Among these, the prognostic nutritional index (PNI) has been widely used, owing to its efficiency, simplicity, and convenience in assessing the preoperative condition and in predicting the surgical risk for gastrointestinal malignancy patients (Onodera et al. 1984). It is calculated using only two values: the serum albumin concentration and lymphocyte count in the peripheral blood, data that can be easily obtained. PNI was first suggested to be a nutritional index and a predictor of surgical risk by Buzby et al. in 1980, and this was corroborated by Onodera et al. in 1984 (Buzby et al. 1980; Onodera et al. 1984). Since then, it has been further investigated, with numerous recent studies having shown that a low PNI is an independent adverse prognostic factor for short-term post-operative complications and long-term outcomes in many different kinds of cancers, such as gastric cancer, colorectal cancer, and oesophageal carcinoma (Nozoe et al. 2002, 2012; Watanabe et al. 2012). However, most of these studies were conducted with relatively small sample sizes and thus carry limited illustrative power to draw convincing conclusions. Accordingly, we conducted a meta-analysis to systematically review the published studies on the association between PNI and the prognoses of various cancers in order to provide more potent evidence to confirm the independent prognostic role of PNI in cancer. To our knowledge, this is the first meta-analysis in this field.
Materials and methods Retrieval strategies Studies concerning the relationship between PNI and cancer patient survival were retrieved from PubMed, Medline, and ISI Web of Science. The following MeSH terms and free text words were used: ‘prognostic nutritional index’, ‘PNI’, ‘cancer’, ‘carcinoma’, ‘tumour’, and ‘neoplasm’. The reference lists of the retrieved articles were also reviewed to identify additional relevant papers. To limit publication bias, the search was not restricted by language or publication year. The last date of the literature
13
J Cancer Res Clin Oncol
search was March 20, 2014. The entire retrieval process was performed independently by two authors (Sun KY and Chen SL), and a third person (Xu Jianbo) was consulted to resolve any disagreements. Study selection criteria After the initial literature search, we included studies that met the following selection criteria: (1) investigation of the prognostic value of PNI in any type of cancer; (2) available data for calculating survival estimates, such as odds ratio (OR) or hazard ratio (HR) with 95 % confidence intervals (CIs); (3) availability of full text; and (4) publication in English. Abstracts, meetings, or case reports were excluded. Assessment of paper quality The quality of each paper was assessed as previously described (Steels et al. 2001; Xing et al. 2013). Four main methods were evaluated, including scientific design, laboratory methodology, generalizability, and result analysis. There were four to seven items for each method. Each item was scored as follows: if it was clearly and accurately defined, 2 points; if it was unclear or incomplete, 1 point; and if it was not defined or inadequate, 0 points. The final score was denoted as a percentage ranging from 0 % to 100 %, with a higher value indicating better quality. The evaluation process was performed independently by two authors (Sun KY and Chen SL). Disagreements were solved by consultation with a third person (Xu Jianbo). Data extraction We extracted the data of each individual study in terms of three main aspects: the main characteristics of the studies, including the author, publication year, region, type of cancer analysed, sample size, average age of the study population, cut-off value, surgical condition, outcome measures, and quality score; estimates such as ORs or HRs concerning the prognostic significance of PNI in terms of overall survival (OS), cancer-specific survival (CSS), or post-operative complications; and ORs concerning the association between PNI and clinicopathological features of different types of cancers. Three methods, as defined in previous studies, were used for the extraction of HRs (Parmar et al. 1998; Tierney et al. 2007). The first and most accurate way was to obtain the estimate directly from the original article or to calculate the HR from the O-E statistic and variance if available. If not, the estimate was calculated using the relevant data, such as the number of patients at risk in each group, the number of events, and the log-rank statistics or its P value. However,
J Cancer Res Clin Oncol Fig. 1 Flow diagram of the meta-analysis
when these data were also unavailable, the HR was determined from the Kaplan–Meier survival curves under the assumption that the rate of patients with censored data during the follow-up was constant. We calculated an approximate HR by extracting several survival rates at specified times from survival curves using the Engauge Digitizer version 2.11.
plots were generated for visual inspection to qualitatively assess publication bias, and the Begg-Mazumdar rank correlation test and Egger’s regression asymmetry test were used for quantitatively determining the extent of publication bias (Begg & Mazumdar 1994; Egger et al. 1997). All calculations for the current meta-analysis were performed using Stata statistical software, version 12.0 (Stata, College Station, TX).
Statistical analysis Each HR was obtained as mentioned above and summarised as pooled ORs or HRs. The results were reported as ORs and their 95 % CIs. They were first calculated using the random-effects model to identify heterogeneity. If heterogeneity was not significant, the fixed-effects model (Mantel–Haenszel) was subsequently used (Higgins et al. 2003). Heterogeneity was assessed by forest plots, the inconsistency test (I2), and the chi-square (χ2) test, with P
Review - Clinical Oncology
The prognostic significance of the prognostic nutritional index in cancer: a systematic review and meta‑analysis Kaiyu Sun · Shuling Chen · Jianbo Xu · Guanghua Li · Yulong He
Received: 19 April 2014 / Accepted: 13 May 2014 © Springer-Verlag Berlin Heidelberg 2014
Abstract Purpose The prognostic nutritional index (PNI) is a simple and effective parameter, initially created to evaluate preoperative nutritional conditions and surgical risk. It has been recently been found to be associated with shortand long-term outcomes of various malignancies. We performed a meta-analysis to determine the predictive significance of PNI in cancer, as a mean to assist in determining the optimal surgery timing and in improving the survival of cancer patients. Methods Data were retrieved from PubMed and ISI Web of Science to identify eligible studies. Odds ratios (ORs) and hazard ratios (HRs) were extracted and pooled to explore the relationships of PNI with patient survival and clinicopathological features. Results Fourteen studies with a total of 3,414 participants met the inclusion criteria. Low PNI was associated with poor overall survival (pooled OR 1.80, 95 % confidence interval [CI] 1.59–2.04) and the presence of post-operative complications (pooled OR 2.45, 95 % CI 1.31–4.58) in cancer patients, but not with cancer-specific survival (CSS) (pooled HR 1.81, Kaiyu Sun and Shuling Chen have contributed equally to the work. K. Sun · J. Xu · G. Li · Y. He (*) Division of Gastrointestinal Surgery, The First Affiliated Hospital, Sun Yat-sen University, Guangzhou 510080, People’s Republic of China e-mail: [email protected] K. Sun e-mail: [email protected] S. Chen Division of Gastroenterology, The First Affiliated Hospital, Sun Yat-sen University, Guangzhou 510080, People’s Republic of China
95 % CI 0.94–3.49). PNI was also found to be associated with invasion depth (pooled OR 5.07, 95 % CI 2.34–10.96) and lymph node metastasis (pooled OR 3.70, 95 % CI 2.32– 5.92) in gastric cancer, whereas TNM stage was the only clinicopathological feature associated with PNI in colorectal carcinoma (pooled OR 1.81, 95 % CI 1.24–2.64). Conclusions PNI might be an effective predictive indicator for the prognosis of cancer, especially digestive system carcinomas. Further studies are required to verify the significance of PNI in clinical practice. Keywords Prognostic nutritional index · Cancer · Prognosis · Meta-analysis Abbreviations CIs Confidence intervals CSS Cancer-specific survival HR Hazard ratio HRs Hazard ratios OR Odds ratio ORs Odds ratios OS Overall survival PNI Prognostic nutritional index
Introduction Cancer has always represented a major medical problem and is associated with high morbidity, mortality, and economic burden, despite great improvements in early detection, surgical techniques, chemotherapy, radiotherapy, biological treatment, and multidisciplinary treatment in recent years (Jemal et al. 2011). Surgery remains the most important tool for the treatment of solid tumours, if allowed by the condition of the patients. However, many patients develop
13
post-operative complications or even relapse after surgery. The preoperative conditions of the patients, especially the nutritional and immunological conditions, have been found to be associated with both the post-operative prognosis and the long-term outcomes of malignant tumours (Schwegler et al. 2010). Therefore, it is crucial to identify an effective nutritional and immunological index to predict patient prognosis in order to establish the optimal preoperative medical treatment and to determine the optimal time to operate. Several nutritional and immunological indices to predict cancer patients’ prognoses have been identified (Nozoe et al. 2010; Asher et al. 2011; Chua et al. 2011; Li et al. 2014). Among these, the prognostic nutritional index (PNI) has been widely used, owing to its efficiency, simplicity, and convenience in assessing the preoperative condition and in predicting the surgical risk for gastrointestinal malignancy patients (Onodera et al. 1984). It is calculated using only two values: the serum albumin concentration and lymphocyte count in the peripheral blood, data that can be easily obtained. PNI was first suggested to be a nutritional index and a predictor of surgical risk by Buzby et al. in 1980, and this was corroborated by Onodera et al. in 1984 (Buzby et al. 1980; Onodera et al. 1984). Since then, it has been further investigated, with numerous recent studies having shown that a low PNI is an independent adverse prognostic factor for short-term post-operative complications and long-term outcomes in many different kinds of cancers, such as gastric cancer, colorectal cancer, and oesophageal carcinoma (Nozoe et al. 2002, 2012; Watanabe et al. 2012). However, most of these studies were conducted with relatively small sample sizes and thus carry limited illustrative power to draw convincing conclusions. Accordingly, we conducted a meta-analysis to systematically review the published studies on the association between PNI and the prognoses of various cancers in order to provide more potent evidence to confirm the independent prognostic role of PNI in cancer. To our knowledge, this is the first meta-analysis in this field.
Materials and methods Retrieval strategies Studies concerning the relationship between PNI and cancer patient survival were retrieved from PubMed, Medline, and ISI Web of Science. The following MeSH terms and free text words were used: ‘prognostic nutritional index’, ‘PNI’, ‘cancer’, ‘carcinoma’, ‘tumour’, and ‘neoplasm’. The reference lists of the retrieved articles were also reviewed to identify additional relevant papers. To limit publication bias, the search was not restricted by language or publication year. The last date of the literature
13
J Cancer Res Clin Oncol
search was March 20, 2014. The entire retrieval process was performed independently by two authors (Sun KY and Chen SL), and a third person (Xu Jianbo) was consulted to resolve any disagreements. Study selection criteria After the initial literature search, we included studies that met the following selection criteria: (1) investigation of the prognostic value of PNI in any type of cancer; (2) available data for calculating survival estimates, such as odds ratio (OR) or hazard ratio (HR) with 95 % confidence intervals (CIs); (3) availability of full text; and (4) publication in English. Abstracts, meetings, or case reports were excluded. Assessment of paper quality The quality of each paper was assessed as previously described (Steels et al. 2001; Xing et al. 2013). Four main methods were evaluated, including scientific design, laboratory methodology, generalizability, and result analysis. There were four to seven items for each method. Each item was scored as follows: if it was clearly and accurately defined, 2 points; if it was unclear or incomplete, 1 point; and if it was not defined or inadequate, 0 points. The final score was denoted as a percentage ranging from 0 % to 100 %, with a higher value indicating better quality. The evaluation process was performed independently by two authors (Sun KY and Chen SL). Disagreements were solved by consultation with a third person (Xu Jianbo). Data extraction We extracted the data of each individual study in terms of three main aspects: the main characteristics of the studies, including the author, publication year, region, type of cancer analysed, sample size, average age of the study population, cut-off value, surgical condition, outcome measures, and quality score; estimates such as ORs or HRs concerning the prognostic significance of PNI in terms of overall survival (OS), cancer-specific survival (CSS), or post-operative complications; and ORs concerning the association between PNI and clinicopathological features of different types of cancers. Three methods, as defined in previous studies, were used for the extraction of HRs (Parmar et al. 1998; Tierney et al. 2007). The first and most accurate way was to obtain the estimate directly from the original article or to calculate the HR from the O-E statistic and variance if available. If not, the estimate was calculated using the relevant data, such as the number of patients at risk in each group, the number of events, and the log-rank statistics or its P value. However,
J Cancer Res Clin Oncol Fig. 1 Flow diagram of the meta-analysis
when these data were also unavailable, the HR was determined from the Kaplan–Meier survival curves under the assumption that the rate of patients with censored data during the follow-up was constant. We calculated an approximate HR by extracting several survival rates at specified times from survival curves using the Engauge Digitizer version 2.11.
plots were generated for visual inspection to qualitatively assess publication bias, and the Begg-Mazumdar rank correlation test and Egger’s regression asymmetry test were used for quantitatively determining the extent of publication bias (Begg & Mazumdar 1994; Egger et al. 1997). All calculations for the current meta-analysis were performed using Stata statistical software, version 12.0 (Stata, College Station, TX).
Statistical analysis Each HR was obtained as mentioned above and summarised as pooled ORs or HRs. The results were reported as ORs and their 95 % CIs. They were first calculated using the random-effects model to identify heterogeneity. If heterogeneity was not significant, the fixed-effects model (Mantel–Haenszel) was subsequently used (Higgins et al. 2003). Heterogeneity was assessed by forest plots, the inconsistency test (I2), and the chi-square (χ2) test, with P
