1. Introduction
Periampullary neoplasms are characterized as a heterogeneous group of tumors with different histological lineages but common features, such as anatomical location, diagnosis at advanced stages, and a very poor prognosis despite radical surgery[
1]. This term includes tumors of the head of the pancreas, ampulla of Vater, distal bile duct, and duodenum adjacent to the ampullary area.
The incidence is 1,401,450 new cases diagnosed each year worldwide [
2]. It is the fourth leading cause of cancer death in the USA and is predicted to become the second leading cause of cancer death by 2030[
3]. This disease is associated with a poor prognosis, reflected by a 5-year survival of less than 5% when all stages are grouped together[
4].
Several prognostic factors have been identified for periampullary and pancreatic cancer.
Preoperative Carbohydrate-antigen 19-9 (CA 19.9) levels in the case of adenocarcinomas[
5], the surgical margin after resection, venous and/or arterial vascular involvement, or histopathological features such as the existence of affected locoregional nodes or perineural or lymphovascular invasion, are the most relevant [
6,
7,
8]. Among all these factors, perineural invasion may be the primary independent factor affecting the prognosis of these patients[
9,
10,
11,
12] particularly in patients with relatively favorable pathological features[
13].
In this process, cancer cells spread along the nerves and their sheaths, using the perineural spaces as routes for tumor dissemination[
11]. Although it is more frequent in patients with adenocarcinoma[
14], perineural infiltration may also be present in tumors of other histological origin, such as neuroendocrine tumors [
15].
The identification of perineural invasion prior to surgery may influence the clinical management of these patients. It could facilitate the stratification of risk, allowing for the identification of those patients who would most benefit from neoadjuvant treatments or more aggressive surgical procedures, such as vascular resections. However, perineural involvement is very difficult to detect preoperatively.
There are only a few studies that have analyzed the possible relationship between CA 19.9 and perineural invasion[
16], and their results are inconsistent. Alternatively, some studies have demonstrated the usefulness of preoperative multidetector CT to detect extrapancreatic perineural invasion [
17,
18,
19]. However, these radiological tests are not yet useful to directly detect microscopic perineural invasion, especially in early stages.
On the other hand, in recent years, evidence has been supporting that the systemic inflammatory response could play an important role in relation to the prognosis of different neoplasms[
20]. Within these inflammatory reactants, the neutrophil-lymphocyte ratio (NLR) has been gaining prominence[
21]. Regarding pancreatic cancer, there are many studies that suggest that the presence of high preoperative levels of the NLR is related to a worse prognosis [
21,
22,
23,
24,
25]. However, no studies have been previously published in which a possible relationship between this inflammatory marker and perineural infiltration has been analyzed.
The aim of this study was to investigate whether preoperative levels of NLR and CA 19.9 marker could predict the presence of histopathologically objectified perineural invasion after surgery. We hypothesize that these two simple preoperative laboratory tests could behave as good markers of perineural invasion in patients with periampullary and pancreatic cancer.
2. Methods
2.1. Study Design and Participants
A retrospective, cross-sectional, observational study was conducted on all patients who were operated on for neoplasia of the biliopancreatic junction and pancreatic cancer between January 2013 and August 2023 in our institution. The setting was a tertiary referral center that serves a population of approximately 400,000 inhabitants.
Inclusion criteria: patients with a histopathological diagnosis of malignancy, in whom the neoplasia was resected and who had a pre-surgical complete blood count that made it possible to calculate the RNL index.
Exclusion criteria: Patients with non-malignant neoplasms, non-operated patients, and those with incomplete or unavailable clinical records were excluded. The number and characteristics of the excluded patients were not collected.
This study was approved by the Ethics Committee of the hospital (Code 2024-260-1). The work has been reported in line with the STROBE Statement.
2.2. Management of the Patient
After completing a thorough medical history and physical examination, blood tests including a complete blood count, blood biochemistry, liver function tests, and tumor markers CEA and CA 19.9 were requested. In recent years, a nutritional study of the patient has also been added. All patients underwent preoperative thoraco-abdominal tomography and/or magnetic resonance imaging to exclude distant metastasis, and to determine the local extent and resectability of the neoplasm. Electrocardiogram, and additional tests were also performed based on each patient’s underlying condition.
The patients were then classified according to NCCN guidelines[
26] as resectable, borderline or locally advanced and evaluated by a multidisciplinary tumor committee. In cases of diagnostic doubt and/or for patients referred for neoadjuvant therapy, a fine-needle aspiration cytology was performed under conventional ultrasound or endoscopic ultrasonography guidance.
All patients with resectable neoplasms and no risk factors were submitted for upfront surgery. Neoadjuvant treatment was proposed for those patients whose tumor was classified as borderline, locally advanced or resectable but with biological criteria of poor prognosis. Among the latter, CA 19.9 levels of over 200 U/mL, decision that was implemented for patients operated on in the last years.
The systemic treatment administered was FOLFIRINOX or gemcitabine/abraxane depending on the characteristics of the patient after evaluation by an expert oncologist. In some patients, radiotherapy was recommended after chemotherapy.
In patients initially referred for upfront surgery, those who responded or in whom the disease did not progress after neoadjuvant surgery, a cephalic duodenopancreatectomy with Child type reconstruction was performed. In some patients, total pancreaticoduodenectomy was carried out.
The resection specimens were analyzed by a pathologist to determine the histologic nature of the tumors according to a previously established protocol[
27].
2.3. Variables, Data Collection and Definitions
Data were retrospectively collected from the computerized database of the hospital and all medical records were reviewed. The following variables were analyzed:
Demographic and clinical data: age, sex, body mass index (BMI), comorbidity measured by Charlson Comorbidity Index, and surgical risk according to ASA classification.
Data laboratory: pre-therapy blood count and Neutrophil-to-Lymphocyte ratio. The NLR was obtained by dividing the absolute number of neutrophils by the absolute number of lymphocytes in the blood count. This variable was categorized according to a cut-off point of 2.2 points obtained from the calculation of the Youden index[
28] in our series. Tumoral markers Carbohydrate-antigen 19-9 and Carcinoembryonic antigen (CEA) also were recorded. The reference values for CA 19.9 were 0.00-37 U/mL, and for CEA, they were 0.00-5.00 ng/mL.
Preoperative local staging according to NCCN classification[
26]: Based on radiological findings, tumors were classified preoperatively into resectable, borderline and locally advanced tumors[
26].
Neoadjuvant therapy: The number of patients who received neoadjuvant therapy (chemo and/or radiotherapy) was collected.
Surgical data: Surgical procedure, surgical approach, postoperative complications according to Clavien-Dindo classification[
29]
, and operative mortality were recorded. Operative mortality was defined as any death occurring within 90 days of surgery or any subsequent death that was determined to be a direct consequence of a postoperative complication.
Histopathological features: The histological type, resection margins, and existence of perineural and/or lymphovascular invasion were recorded. The tumors were classified into four histopathologic stages according to the American Joint Committee on Cancer (AJCC) TNM Staging of Pancreatic Cancer (8th ed.)[
30], regardless of the histological type of tumor.
Output measure: Perineural invasion was the variable output. It was defined as the presence of cancer cells along the nerves and/or in the epineural, perineural or endoneural space of the neural sheaths[
31] .
2.4. Statistical Analysis
The statistical software package SPSS 29.0 for Windows (IBM Corporation, Armonk, NY, USA) was used to analyze the data.
Categorical variables were expressed as frequencies and percentages. Continuous quantitative variables were expressed as means and standard deviations (±SD) when the data followed a normal distribution, or medians and interquartile ranges (IQR = 25–75th percentile) when they did not.
Next, a univariate analysis was performed to assess the association between the different predictor variables and perineural infiltration as the output variable. For univariate analysis, the χ2 test or Fisher's exact test was used in the case of categorical variables. For normally distributed numerical variables, Student’s t-test was used to compare means. Mann–Whitney U test was used for non-normal distribution.
Multivariate analysis (logistic regression) was performed to detect independent prognostic factors related to perineural infiltration. The most clinically revealing variables that were statistically significant (p<0.05) in the univariate analysis were introduced into the model.
Receiver operative characteristic (ROC) curve analyses were performed to test the discriminatory ability of the NLR and CA 19.9 variables in order to predict perineural infiltration. ROC curves were constructed and the areas under the curve (AUROC) were compared.
For basal NLR, the Youden index (sensitivity + specificity-1)[
28] was used to select a threshold for estimating sensitivity and specificity. In the case of CA 19.9, the cut-off point used was 37 U/mL, the laboratory's reference value. Sensitivity, specificity, positive predictive value (PPV), and negative predictive value (PPN) were calculated for both variables.
A significance level of p<0.05 was considered and the odds ratio (CI 95%) was used as measure of the magnitude of the association.
4. Discussion
There is a growing interest in identifying accessible, easy-to-evaluate, and cost-effective prognostic markers in the management of certain tumors, such as periampullary and pancreatic cancer. In these cases, early detection of risk factors for poor outcome is considered essential in order to optimally plan the sequence of application of the different available therapeutic modalities and the radicality of surgery.
This does not usually cause problems in the case of borderline or locally advanced tumors, which are easily detectable with preoperative radiological studies. But it may present difficulties in early-stage tumors. Perineural infiltration, an ominous sign of poor evolution and which may already be present in early stages[
13], cannot be diagnosed until the resection specimen is studied histologically after surgery.
Among the classic preoperative prognostic factors, CA 19.9 is the main biomarker predicting survival and resectability in biliopancreatic neoplasms. Patients who have elevated CA 19-9 levels at diagnosis are biologically borderline resectable regardless of anatomic resectability, and neoadjuvant systemic therapy is recommended[
5].
In our study, in an attempt to detect perineural invasion early, we demonstrate that both the tumor marker CA 19-9, and especially the inflammatory marker LNR, are good predictors of the presence of perineural invasion, and may be useful as an indirect sign of poor prognosis.
Multivariate analysis showed that an NLR above level 2.20 was an independent prognostic factor for perineural involvement in the surgical specimen. A patient with a NLR > 2.20 was 95% more likely to have perineural infiltration. Something similar happened with CA 19.9. An elevation of this marker indicated a 67% likelihood of perineural infiltration.
However, up to 10-15% of patients (Lewis negative phenotype) may fail to produce CA 19.9[
32], and its use can be limited by false negatives and false positives, especially in the presence of obstructive jaundice[
33].
Moreover, we found a group of 24 patients (19.5% of the total sample) with normal CA 19.9 levels and high NLR levels. Of these, 75% had perineural infiltration. Therefore, it is in these situations when the NLR would become more relevant. If our results are confirmed in other studies, patients diagnosed with pancreatic cancer with high levels of NLR and a normal CA 19.9 before surgery, should be referred to the oncologist for neoadjuvant therapy.
Inflammation has been long recognized as a key aspect in the development and prognosis of different neoplasms[
34]. Neutrophils are promoters of angiogenesis and may play a role in tumor cell proliferation and the development of metastases[
35]. Signals emanating from cancer cells educate neutrophils to facilitate support networks that lead to its expansive spread. These functions can occur locally in or around the tumor microenvironment, as well as systemically in distant organs[
36]. In this sense, NLR should be considered a good systemic inflammatory marker.
In fact, a close relationship between preoperative NLR and long-term survival in pancreatic cancer has been demonstrated[
21,
22,
23,
24,
25]. Only few authors have not found this association in their series[
37,
38]. Other inflammatory prognostic markers such as C-reactive protein, alone or in combination with albumin[
39,
40] or certain scores such as the Systemic Immune-inflammatory index[
41] have also been used. This aspect was not analyzed in our study.
The role of perineural invasion in cancer biology has been underestimated, often considered a passive process with less significance compared to lymphatic and hematogenous dissemination. However, new evidence shows neural regulation in cancer and cancer-induced axonogenesis. It is being emphasized that nerves are emerging regulators of cancer initiation, progression, and metastasis[
42]. Some authors[
12] even suggest that curative pancreatic tumor resection with long-term survival could only be achieved in perineural invasion negative patients.
Perineural infiltration is a complex process that depends on a combination of biological, molecular, and tumor microenvironment factors[
43].
The periampullary area and the head of the pancreas have extensive innervation from the autonomic nervous system through the celiac plexus and the superior mesenteric artery[
44]. Due to the high neurotropism of tumor cells in pancreatic adenocarcinoma, these plexuses represent a pathway for dissemination and a possible cause of local recurrence and lymphatic metastases[
45]. Thus, perineural infiltration is detected in almost 75% of early-stage pancreatic adenocarcinomas, suggesting that it represents an early sign of tumor progression[
31,
46].
In our series, we have included neuroendocrine and some very poorly differentiated tumors. One third of our patients with neuroendocrine tumors had perineural infiltration. We know that perineural infiltration may be present to varying degrees in neuroendocrine tumors[
15], and that it also behaves as an important predictor of survival[
47].
In this context, we believe that the location of the tumor may be even more relevant than the histological type of neoplasia in predicting perineural invasion. The complex neural network in the periampullary region provides tumors with proximity to nerves, thereby increasing the risk of neural invasion. It has been observed that there is a significant difference in the rate of neural invasion between intrahepatic and extrahepatic anatomical locations, despite both belonging to the category of cholangiocarcinomas.
Two meta-analyses have been performed[
10,
48] to evaluate the
influence of neural invasion on survival and tumor recurrence in pancreatic ductal adenocarcinomas. They concluded that PNI appears to be an independent prognostic factor for overall survival, disease-free survival and progression-free survival in these neoplasms. The authors emphasized that PNI should be increasingly considered in patient stratification and in the development of novel therapeutic algorithms.
More recently, other series have been published with a large number of patients with the aim of evaluating the impact of perineural invasion in patients who had undergone pancreatic cancer surgery.
Felsenstein et al[
12] found that, out of 571 patients, 93% exhibited tumors with perineural invasion. R0 resection rate was 62.9%., and PNI was observed in the majority of R0 resected patients (91.4%). PNI was associated with worse disease-free and long-term survival. The potential impact of different variables on perineural invasion was also assessed. This variable correlated with advanced tumor stage and lymphatic dissemination in univariate analysis, but in multivariate analysis, only lymph node metastasis was a significant predictor of PNI.
Crippa et al[
9] also published the results of a multicenter study assessing the clinical impact of perineural invasion in pancreatic cancer. PNI was found in 87% of 778 patients included in the study. R0 resection was achieved in 62.9%, and perineural invasion was detected in the majority of R0 tumors (91%). PNI was identified in multivariate analysis as a relevant prognostic factor for both tumor recurrence and survival after pancreatoduodenectomy.
In our serie, PNI was detected in 70% of the patients. R0 resection rate was 69.1%, and perineural invasion was observed in only 57.9% of R0 tumors. These differences could be explained by the type of patients included in our series: the majority were initially considered resectable. More complex patients were probably included in the two previously mentioned studies.
Schouten et al[
13], in a nationwide observational cohort study of 1630 Dutch patients, confirmed the previous finding that perineural invasion seems to be an important prognostic factor in patients with radically resected (R0) and node-negative (pN0) pancreatic cancer. They concluded that PNI is strongly associated with worse survival, particularly in resected patients with relatively favorable pathological features.
It is challenging to identify perineural invasion before surgery. In our series, perineural infiltration was statistically significantly associated with NLR, CA 19.9, CEA, histological type, tumor stage and lymph node involvement.
We have not found in the literature consulted any reference relating LNR to the presence of perineural invasion. Regarding CA 19.9, our results are comparable with the reported rates of Wang et al[
16]. However, CA 19.9 was not related to PNI in the series of Felsenstein et al[
12], Crippa et al[
9], and Zou et al[
49]. We note that the distribution of the CA 19.9 variable was highly skewed and that a logarithmic transformation had to be performed to analyze its significance.
This marker is not applicable to detect PNI in neuroendocrine tumors. In our series, NLR values were also not associated with the presence of PNI. In these cases, higher tumor grades seem to be more associated with the existence of PNI[
50].
We confirm, as did other authors, that perineural infiltration was most frequently observed in ductal adenocarcinomas[
14], followed by biliary adenocarcinomas, undifferentiated carcinomas, neuroendocrine tumors, and intestinal adenocarcinomas. These differences were statistically significant. Also, a greater number of patients with PNI were detected in the more advanced stages of the disease.
In an attempt to improve the preoperative diagnosis of PNI, some alternatives have been proposed. Zhang et al[
51] suggest that PNI could be diagnosed preoperatively by evaluating abdominal enhanced MSCT images with high accuracy, combined with serum tumor marker levels. In this study, classification of fat tissue around celiac trunk incorporated with tumor size, CA19-9, CEA, age and sex, showed the highest AUC as 0.939, with specificity of 95.0% and sensitivity of 90.4%. Zou et al[
49] developed a nomogram using preoperative total bilirubin, preoperative blood glucose, and preoperative CA19-9. It achieved good AUC values of 0.753 and 0.737 in predicting PNI in training and validation cohorts, respectively. Other authors[
18] argue that certain CT findings provides sufficient diagnostic information to detect PNI in patients with resectable tumors of the pancreatic head before surgery.
Regarding the predictive capability of the preoperative markers under consideration, the area under the ROC curve obtained was better for the NLR (0.72) than for the CA 19.9 marker (0.67). In this latter case, AUROC was lower than expected. Nevertheless, with both markers, a PPV greater than 80% was achieved. This implies that 80% of patients with NLR levels >2.2 and/or CA 19.9 levels >37 will have perineural infiltration.
The cut-off point for NLR is highly debated. Forget et al[
52], identified that the normal NLR values in an adult, non-geriatric, population in good health are between 0.78 and 3.53. Here we used 2.2 as cut-off point, which was the value that maximized the sum of sensitivity and specificity in our series, according to the Youden index. Interestingly, it was the same NLR cutoff point used by Hasegawa et al[
53] in their study. Inoue et al[
54] used a NLR cutoff point of 2. However, in most of the series included in the systematic review by Stevens et al[
39], cut-off points around 5 were used.
Our study has some limitations. First, it is a single-center, retrospective study with a limited number of patients. Additionally, other potential confounding factors such as nutritional status or symptom duration have not been analyzed. Therefore, these results should be confirmed in larger cohorts in prospective multicenter studies. Nevertheless, these findings emphasize and demonstrate the value of an easily accessible inflammatory marker for the clinician, the NLR, in patients with periampullary and pancreatic cancer, especially in patients with normal CA19.9.