Article Info

Author Affiliations: HPB Surgery, Mater Hospital, Crumlin Road, Belfast, Antrim, BT14 6AB, UK (Ahmad J and Grimes N); Department of Surgery, Kettering General Hospital, Rothwell Road, Kettering, Northamptonshire, NN16 8UZ, UK (Farid S); Department of Surgery, Belfast City Hospital, Lisburn Road, Belfast, Antrim, BT9 7AB, UK (Morris-Stiff G)

Acknowledgements: We thank Chris Cardwell (Senior Biomedical statistician, Queen's University Belfast) for his advice on statistical analysis and Richard Fallis (Senior Librarian, Queen's University Belfast) for his help with database searching.

BACKGROUND: Various scoring systems based on assessment of the systemic inflammatory response help assessing the prognosis of patients with pancreatic ductal adenocarcinoma. In the present systematic review we evaluated the validity of four pre-intervention scoring systems: Glasgow prognostic score (GPS) and its modified version (mGPS), platelet lymphocyte ratio (PLR), neutrophil lymphocyte ratio (NLR), and prognostic nutrition index (PNI).

DATA SOURCES: MOOSE guidelines were followed and EMBASE and MEDLINE databases were searched for all published studies until September 2013 using comprehensive text word and MeSH terms. All identified studies were analyzed, and relevant studies were included in the systematic review.

RESULTS: Six studies were identified for GPS/mGPS with 3 reporting statistical significance for GPS/mGPS on both univariate analysis (UVA) and multivariate analysis (MVA). Two studies suggested prognostic significance on UVA but not MVA, and in the final study UVA failed to show significance. Eleven studies evaluated the prognostic value of NLR. Six of them reported prognostic significance for NLR on UVA that persisted at MVA in 4 studies, and in the remaining 2 studies NLR was the only significant factor on UVA. In the remaining 5 studies, all in patients undergoing resection, there was no significance on UVA. Seven studies evaluated PLR, with only one study demonstrated its prognostic significance on both UVA and MVA, the rest did not show the significance on UVA. Of the two studies identified for PNI, one demonstrated a statistically significant difference in survival on both UVA and MVA, and the other reported no significance for PNI on UVA.

CONCLUSIONS: Both GPS/mGPS and NLR may be useful but further better-designed studies are required to confirm their value. PLR might be little useful, and there are at present inadequate data to assess the prognostic value of PNI. At present, no scoring system is reliable enough to be accepted into routine use for the prognosis of patients with pancreatic ductal adenocarcinoma.

Pancreatic ductal adenocarcinoma (PDAC) has a poor prognosis with a worldwide annual death toll of approximately 227 000.^[1] Presently, surgical resection is the only chance of cure for these patients, but this is possible in only 20% of cases.^[2] Surgery itself is a major undertaking and has a high rate of peri-operative morbidity.^[3] Furthermore, despite the presence of favorable prognostic features (small tumor, negative resection margins, absence of lymph node involvement and no lymphovascular invasion), the 5-year survival for patients with pancreatic cancer is 18%.^[4]

It was shown that outcomes are improving in well-selected patients mainly because of the improvement of peri- and post-operative care.^[5] Modern cross-sectional imaging improved our ability to assess resectability of pancreatic cancer in the pre-operative setting; however, tumor behavior remains difficult to assess before excision and a detailed histological examination. There is therefore the need for a robust marker that could be used in the prognostic stratification of patients with PDAC prior to treatment.

Scoring systems based on assessment of the systemic inflammatory response have been shown to be useful pre-intervention indicators of prognosis in various malignancies.^{[6, 7]} Those assessed in PDAC include: Glasgow prognostic score (GPS) including its modified version (mGPS); platelet lymphocyte ratio (PLR); neutrophil lymphocyte ratio (NLR); and prognostic nutrition index (PNI). However, none of them have been universally accepted and so their role in patients with PDAC remains unclear.

A thorough literature search was performed to identify all studies that examined the prognostic value of one or more of the four scoring systems. All studies published until the first week of September 2013 were evaluated. EMBASE and MEDLINE databases were searched using a comprehensive text word and MeSH terms (exp pancreatic neoplasms/, Glasgow prognostic score*, modified Glasgow prognostic score*, GPS*, mGPS*, platelet lymphocyte ratio*, PLR*, neutrophil lymphocyte ratio*, NLR*, prognostic nutrition index*, Onodera's prognostic nutrition index* and PNI*). Google Scholar was used to identify additional manuscripts. A bibliography of included articles was hand searched for relevant additions. Two investigators (Ahmad J and Grimes N) independently searched, scored and analyzed all data, and consensus was sought where discrepancies existed.

As all studies were observational, our search protocol was based on meta-analysis of observational studies in epidemiology (MOOSE) guidelines.^[8] All papers indexed in the English language were included. Case reports or case series, unpublished abstracts, proceedings of meetings, and animal studies were excluded. After removal of duplicates and irrelevant articles, abstracts were screened for further exclusions. Full text articles were then read before inclusion in the analysis. Critical appraisal skills programme (CASP)^[9] guidelines were used to appraise the quality of included papers. Data were evaluated for meta-analysis and statistical advice was sought from a bio-statistician.

The results from searching the EMBASE and MEDLINE databases are presented in Table 1. Amongst the included studies, six focused on the prognostic value of GPS and mGPS, eleven assessed NLR, seven investigated the value of PLR, and two evaluated the predictive role of PNI in PDAC. PRISMA^[10] flow diagrams are shown in Figs. 1-4 for each of the 4 scoring systems, and details of the systematic review are presented in Tables 2-6.

Three studies evaluated mGPS^[11-13] and three assessed the prognostic value of GPS.^[14-16] As the core elements of the score are similar, they were grouped together. Two of these studies focused on patients who underwent surgery but two concentrated on patients with unresectable or metastatic disease. The other 2 studies presented data for both resected and unresected disease. Three of the 6 studies indicated that GPS/mGPS was capable of providing statistically significant prognostic information on both univariate analysis (UVA) and multivariate analysis (MVA). Two studies showed significance on UVA but not MVA and one study showed no prognostic significance on UVA. The 3 studies providing data on median survival all showed significance on MVA, and the median survival was 2.6-5.1 times greater in patients with a score of 0 than in those with a score of 2.

Eleven studies evaluated NLR in patients with PDAC.^{[12, 13, 15, 17-24]} Seven of these studies examined patients undergoing resection alone, three evaluated unresectable and resected patients and one looked at a population not undergoing resection. When the patients were classified according to NLR, differences in outcomes were shown to be statistically significant in 6 studies on UVA. In 4 of these studies, significance was seen on MVA, and in the remaining 2 studies, NLR was the only variable of prognostic significance on UVA and so MVA was not required. In the remaining 5 studies, which assessed patients undergoing resection, NLR was not identified as a significant variable on UVA. In the studies on the prognostic value of NLR, there was a 2.0-4.3 fold survival difference in low and high values of NLR, corresponding to an NLR <5 versus >5 in all but one study, the final evaluation using a NLR of 4 as a cut off. Indeed, only Jamieson et al^[22] indicated a negative impact on a low NLR as the remaining 4 studies showed an improved survival in the NLR <5 group (a 1.2-1.8 fold increase survival).

Seven studies aimed to examine the prognostic value of PLR^{[12, 13, 17, 19, 22-24]} in relation to outcomes. Four of them focused on the prognostic value of PLR in patients who underwent surgery with curative intent and three on that in both resected and unresectable cases. Only one study^[24] found that PLR was a statistically significant scoring system on both UVA and MVA, the other 6 studies reported no significance on UVA. Despite heterogeneity in the values assigned to PLR for assessment and also the lack of statistical significance, comparison of low and high PLRs indicated that survival was significantly better in the presence of a low PLR with a 1.3-3.4 fold increase in median survival.

In two studies identified for PNI,^{[13, 25]} one demonstrated a difference in survival on UVA and MVA, and the other reported no significance for PNI on UVA. Kanda et al^[25] indicated a 1.7-fold increase in survival when the PNI was <45.

All data were assessed for inclusion in a meta-analysis. Unfortunately, the heterogeneity was such that a reliable meta-analysis was not possible for a variety of reasons. These included the fact that the exact number of cases was not known in some studies and that different outcome measures were used. Furthermore, whilst some studies concentrated on resectable or unresectable tumors, or indeed provided raw data for these subgroups, other studies combined patients undergoing resection and those with unresectable disease into a single group. The equation used to calculate the prognostic scores was also not consistent across the board. Following discussion with a medical statistician, it was therefore decided that a meta-analysis would not be appropriate.

The prognosis of any cancer depends not only on tumor biology, but also on patient related factors. The systemic inflammatory response is acknowledged to be an important consideration in many kinds of tumors including pancreatic cancer.^[26] Chronic inflammation not only has a direct causal relationship with tumorigenesis, but malignancies themselves trigger an inflammatory response that leads to many deleterious effects of the malignant process including cachexia.^[27] Cancer-related inflammation causes suppression of anti-tumor immunity through recruitment of regulatory T cells and activation of chemokines encouraging tumor growth and metastasis. It has therefore been hypothesized that tumor growth is directly proportional to the degree of inflammation.^[19]

The scoring systems assessed in this review were based on elements of the inflammatory response namely the neutrophil count, platelet count, lymphocyte count, albumin and C-reactive protein (CRP), in different combinations or equations.

Many cancers have been shown to incite a systemic inflammatory response as evidenced by an elevated CRP.^[28] There is an increasing evidence that patients with a significant host inflammatory response have a poorer prognosis, and host inflammatory response imparts an independent prognostic value in relation to outcomes.^[29-33] The mechanism of how the systemic inflammation affects survival in cancer patients is not clearly understood. It is however agreed that the systemic inflammatory response leads to the release of pro-inflammatory cytokines such as interleukin(IL)-6, and growth factors that not only stimulate tumor growth but also induce profound catabolic effects on host metabolism.^[14]

Neutrophilia is commonly seen in cancer patients and is a multifactorial process. Neoplastic processes produce myeloid growth factors as part of a paraneoplastic syndrome, and these trigger excessive neutrophil production. In addition, tumor cells release granulocyte colony-stimulating factors that cause neutrophilia by stimulating the bone marrow.^{[19, 34, 35]} IL-6 and tumor necrosis factor may also be involved in cancer-related inflammation leading to an elevation in neutrophil counts.^[36]

Lymphocytopenia is also a significant finding in patients with pancreatic cancer^[37] and is an indicator of immunosuppression present at a local and systemic level.^[38] Lymphocyte function is also believed to be poor due to the release of inhibitory transforming growth factor-beta and IL-10.^[39]

Thrombocytosis is considered a poor prognostic feature in patients with advanced pancreatic cancer.^[40] The condition is probably related to the stimulation of megakaryocytes by pro-inflammatory mediators (mainly IL-1, IL-2 and IL-6).^[41]

Malnutrition has deleterious effects on the outcome of any major surgery.^{[42, 43]} Patients with pancreatic cancer are usually malnourished, hence immunocompromised at presentation.^[44] This leads to a higher incidence of post-operative complications. Moreover, impaired host immunity can contribute to tumor growth.^[45] As an indicator of malnutrition, albumin has a half-life of around 20 days and reflects a mid or long-term nutritional status.^[46] Its levels have been correlated with the incidence of post-operative complications and survival in gastrointestinal cancer.^[47-49]

Forrest et al^[50] first described GPS as an objective combination of CRP and serum albumin. Patients with an elevated CRP (>10 mg/L) and low albumin (<35 g/L) were given a score of 2, patients with either elevated CRP or low albumin score 1, and patients with neither of these biochemical abnormalities score 0. It was observed that patients with a GPS of 2 had particularly poor prognosis as compared with those with a GPS of 1 or 0. Further investigation showed that low albumin without elevated CRP is rare and hypoalbuminemia on its own is of no prognostic value. The resultant mGPS^[51] thus scores zero in patients with isolated low albumin. It is noteworthy that either combined or assessed individually, both have shown to be independent prognostic factors in a variety of cancers.^{[14, 50, 52]}

In the current systematic review, studies evaluating GPS/mGPS were grouped together as modification of the score does not change the best (0) or worst (2) score and these are the groups that led to the GPS showing statistical significance in the original study. We noted that of 6 studies looking at GPS/mGPS, 3 showed significance on MVA.

NLR is calculated from the differential count by dividing the absolute neutrophil count with the absolute lymphocyte count. The NLR >5 is considered significant and also a poor prognostic marker.^[53] Although neutrophilia and lymphocytopenia are common findings in patients with PDAC, there is no universal evidence for the prognostic role because 6 studies demonstrated that NLR was an independent prognostic indicator and 5 showed no prognostic role for NLR.

Proctor et al^[54] reported the use of a variant of NLR that they termed derived NLR (dNLR). They found that many cancer databases did not hold data on lymphocyte counts and so approximated the lymphocyte count by subtracting the neutrophil count from the total white cell count. In a study of 27 031 patients with a variety of cancers (but not specifically PDAC), they suggested the 2 indices had comparable prognostic values. Szkandera et al^[55] investigated the validity of the 2 indices in a cohort of PDAC patients and found that pre-treatment dNLR can be considered as a promising independent prognostic parameter in these patients. The dNLR may prove useful clinically. However, given the variation between optimal values for NLR and dNLR, further studies are required to assess this predictive score.

PLR was calculated in a similar manner as NLR but with the substitution of platelets for neutrophil count. There was some degree of discrepancy in defining a significant PLR, however, most studies stratified patients into three groups with a PLR <150 indicating good, PLR 150-300 moderate, and PLR >300 poor prognosis. Only one study suggested the clinical usefulness of PLR in this systematic review.

PNI^[56] was calculated using the formula: 10×albumin (g/dL)+0.005×total lymphocyte count (per mm³). A PNI >45 is defined as normal whilst a PNI of <45 scores 1 and is considered an independent marker of poor prognosis in cancer patients. Both the included studies in our review used the same cut off and were therefore homogenous; however one showed statistical significance and the other did not.

There are several potential reasons for the lack of robustness of any of the 4 scores assessed, probably the main one being that patient groups were heterogeneous. Indeed this was the most significant factor preventing a statistical analysis of data to reach to significance. The sample sizes for patients some of the subgroups analyzed were very small in many of the studies too and thus type II error was thus likely.

The rationale of these prognostic scoring systems is to assist clinicians who adopt a more individualized approach to patient care. These scoring systems on their own should not be used to deny patients a chance of a possible curative resection but may be useful, in conjunction with individual surgeon/institution data in providing informed consent to patients prior to a major resection such as pancreatectomy. However, if robust, they could possibly identify aggressive tumor behavior in the pre-operative setting, indicating a need for staging laparoscopy or consideration of neo-adjuvant therapy. If reliable they could also serve as a useful monitoring tool after successful resection. The scoring systems could also be used to stratify patients in clinical trials and may have a role in measuring response to adjutant treatment. They could also be used to select patients for aggressive palliative chemotherapy. Therefore a good prognostic scoring system has great clinical potential.

The major limitation of this systematic review is that the included evidence is from retrospective cohort studies, hence the potential of bias and confounding factors exists. This is a well-known problem for systematic reviews of prognostic scores. Moreover, significant heterogeneity of included studies affects the power of this systematic review. The only means of improving the heterogeneity is to pool all the raw data and analyze as one, which is a testing task in evaluating retrospective cohort studies.

In conclusion, this systematic review has shown that at present, there are inadequate data to support any of the assessed scores as a prognostic indicator for patients with PDAC. The evidence for the use of both GPS/mGPS and NLR suggests that they may be of some value but well designed studies are required before they are being widely introduced. The data for PLR cannot currently support its use and there are inadequate data to assess the role of PNI as a prognostic tool in PDAC.

3 Richter A, Niedergethmann M, Sturm JW, Lorenz D, Post S, Trede M. Long-term results of partial pancreaticoduodenectomy for ductal adenocarcinoma of the pancreatic head: 25-year experience. World J Surg 2003;27:324-329. PMID: 12607060

8 Stroup DF, Berlin JA, Morton SC, Olkin I, Williamson GD, Rennie D, et al. Meta-analysis of observational studies in epidemiology: a proposal for reporting. Meta-analysis Of Observational Studies in Epidemiology (MOOSE) group. JAMA 2000;283:2008-2012. PMID: 10789670

11 Jamieson NB, Mohamed M, Oien KA, Foulis AK, Dickson EJ, Imrie CW, et al. The relationship between tumor inflammatory cell infiltrate and outcome in patients with pancreatic ductal adenocarcinoma. Ann Surg Oncol 2012;19:3581-3590. PMID: 22555345

12 Stotz M, Gerger A, Eisner F, Szkandera J, Loibner H, Ress AL, et al. Increased neutrophil-lymphocyte ratio is a poor prognostic factor in patients with primary operable and inoperable pancreatic cancer. Br J Cancer 2013;109:416-421. PMID: 23799847

15 La Torre M, Nigri G, Cavallini M, Mercantini P, Ziparo V, Ramacciato G. The glasgow prognostic score as a predictor of survival in patients with potentially resectable pancreatic adenocarcinoma. Ann Surg Oncol 2012;19:2917-2923. PMID: 22488099

17 Aliustaoglu M, Bilici A, Seker M, Dane F, Gocun M, Konya V, et al. The association of pre-treatment peripheral blood markers with survival in patients with pancreatic cancer. Hepatogastroenterology 2010;57:640-645. PMID: 20698242

19 Bhatti I, Peacock O, Lloyd G, Larvin M, Hall RI. Preoperative hematologic markers as independent predictors of prognosis in resected pancreatic ductal adenocarcinoma: neutrophil-lymphocyte versus platelet-lymphocyte ratio. Am J Surg 2010;200:197-203. PMID: 20122680

21 Garcea G, Ladwa N, Neal CP, Metcalfe MS, Dennison AR, Berry DP. Preoperative neutrophil-to-lymphocyte ratio (NLR) is associated with reduced disease-free survival following curative resection of pancreatic adenocarcinoma. World J Surg 2011;35:868-872. PMID: 21312035

22 Jamieson NB, Denley SM, Logue J, MacKenzie DJ, Foulis AK, Dickson EJ, et al. A prospective comparison of the prognostic value of tumor- and patient-related factors in patients undergoing potentially curative surgery for pancreatic ductal adenocarcinoma. Ann Surg Oncol 2011;18:2318-2328. PMID: 21267785

23 Sanjay P, de Figueiredo RS, Leaver H, Ogston S, Kulli C, Polignano FM, et al. Preoperative serum C-reactive protein levels and post-operative lymph node ratio are important predictors of survival after pancreaticoduodenectomy for pancreatic ductal adenocarcinoma. JOP 2012;13:199-204. PMID: 22406601

24 Smith RA, Bosonnet L, Raraty M, Sutton R, Neoptolemos JP, Campbell F, et al. Preoperative platelet-lymphocyte ratio is an independent significant prognostic marker in resected pancreatic ductal adenocarcinoma. Am J Surg 2009;197:466-472. PMID: 18639229

29 Chen HH, Chen IH, Liao CT, Wei FC, Lee LY, Huang SF. Preoperative circulating C-reactive protein levels predict pathological aggressiveness in oral squamous cell carcinoma: a retrospective clinical study. Clin Otolaryngol 2011;36:147-153. PMID: 21332670

33 Jamieson NB, Glen P, McMillan DC, McKay CJ, Foulis AK, Carter R, et al. Systemic inflammatory response predicts outcome in patients undergoing resection for ductal adenocarcinoma head of pancreas. Br J Cancer 2005;92:21-23. PMID: 15597096

34 Donskov F, Hokland M, Marcussen N, Torp Madsen HH, von der Maase H. Monocytes and neutrophils as 'bad guys' for the outcome of interleukin-2 with and without histamine in metastatic renal cell carcinoma--results from a randomised phase II trial. Br J Cancer 2006;94:218-226. PMID: 16434984

36 Ulich TR, del Castillo J, Keys M, Granger GA, Ni RX. Kinetics and mechanisms of recombinant human interleukin 1 and tumor necrosis factor-alpha-induced changes in circulating numbers of neutrophils and lymphocytes. J Immunol 1987;139:3406-3415. PMID: 3316383

39 Bellone G, Turletti A, Artusio E, Mareschi K, Carbone A, Tibaudi D, et al. Tumor-associated transforming growth factor-beta and interleukin-10 contribute to a systemic Th2 immune phenotype in pancreatic carcinoma patients. Am J Pathol 1999;155:537-547. PMID: 10433946

40 Suzuki K, Aiura K, Kitagou M, Hoshimoto S, Takahashi S, Ueda M, et al. Platelets counts closely correlate with the disease-free survival interval of pancreatic cancer patients. Hepatogastroenterology 2004;51:847-853. PMID: 15143932

41 Alexandrakis MG, Passam FH, Moschandrea IA, Christophoridou AV, Pappa CA, Coulocheri SA, et al. Levels of serum cytokines and acute phase proteins in patients with essential and cancer-related thrombocytosis. Am J Clin Oncol 2003;26:135-140. PMID: 12714883

43 Schwegler I, von Holzen A, Gutzwiller JP, Schlumpf R, Mühlebach S, Stanga Z. Nutritional risk is a clinical predictor of postoperative mortality and morbidity in surgery for colorectal cancer. Br J Surg 2010;97:92-97. PMID: 20013933

47 Lohsiriwat V, Lohsiriwat D, Boonnuch W, Chinswangwatanakul V, Akaraviputh T, Lert-Akayamanee N. Pre-operative hypoalbuminemia is a major risk factor for postoperative complications following rectal cancer surgery. World J Gastroenterol 2008;14:1248-1251. PMID: 18300352

48 Ryan AM, Hearty A, Prichard RS, Cunningham A, Rowley SP, Reynolds JV. Association of hypoalbuminemia on the first postoperative day and complications following esophagectomy. J Gastrointest Surg 2007;11:1355-1360. PMID: 17682826

49 Greenblatt DY, Kelly KJ, Rajamanickam V, Wan Y, Hanson T, Rettammel R, et al. Preoperative factors predict perioperative morbidity and mortality after pancreaticoduodenectomy. Ann Surg Oncol 2011;18:2126-2135. PMID: 21336514

50 Forrest LM, McMillan DC, McArdle CS, Angerson WJ, Dunlop DJ. Comparison of an inflammation-based prognostic score (GPS) with performance status (ECOG) in patients receiving platinum-based chemotherapy for inoperable non-small-cell lung cancer. Br J Cancer 2004;90:1704-1706. PMID: 15150622

51 McMillan DC, Crozier JE, Canna K, Angerson WJ, McArdle CS. Evaluation of an inflammation-based prognostic score (GPS) in patients undergoing resection for colon and rectal cancer. Int J Colorectal Dis 2007;22:881-886. PMID: 17245566

53 Gomez D, Farid S, Malik HZ, Young AL, Toogood GJ, Lodge JP, et al. Preoperative neutrophil-to-lymphocyte ratio as a prognostic predictor after curative resection for hepatocellular carcinoma. World J Surg 2008;32:1757-1762. PMID: 18340479

55 Szkandera J, Stotz M, Eisner F, Absenger G, Stojakovic T, Samonigg H, et al. External validation of the derived neutrophil to lymphocyte ratio as a prognostic marker on a large cohort of pancreatic cancer patients. PLoS One 2013;8:e78225. PMID: 24223776