Scoring of tumour response after neoadjuvant therapy in resected pancreatic cancer: systematic review.

BACKGROUND: Preoperative chemo(radio)therapy is used increasingly in pancreatic cancer. Histological evaluation of the tumour response provides information on the efficacy of preoperative treatment and is used to determine prognosis and guide decisions on adjuvant treatment. This systematic review aimed to provide an overview of the current evidence on tumour response scoring systems in pancreatic cancer. METHODS: Studies reporting on the assessment of resected pancreatic ductal adenocarcinoma following neoadjuvant chemo(radio)therapy were searched using PubMed and EMBASE. All original studies reporting on histological tumour response in relation to clinical outcome (survival, recurrence-free survival) or interobserver agreement were eligible for inclusion. This systematic review followed the PRISMA guidelines. RESULTS: The literature search yielded 1453 studies of which 25 met the eligibility criteria, revealing 13 unique scoring systems. The most frequently investigated tumour response scoring systems were the College of American Pathologists system, Evans scoring system, and MD Anderson Cancer Center system, investigated 11, 9 and 5 times respectively. Although six studies reported a survival difference between the different grades of these three systems, the reported outcomes were often inconsistent. In addition, 12 of the 25 studies did not report on crucial aspects of pathological examination, such as the method of dissection, sampling approach, and amount of sampling. CONCLUSION: Numerous scoring systems for the evaluation of tumour response after preoperative chemo(radio)therapy in pancreatic cancer exist, but comparative studies are lacking. More comparative data are needed on the interobserver variability and prognostic significance of the various scoring systems before best practice can be established.

The ageing pancreas: a systematic review of the evidence and analysis of the consequences.

Senior people constitute the fastest growing segment of the population. The elderly are at risk for malnutrition, thought to be caused by reduced food intake or involution of the physiological capacity of the GI tract. Age-related changes are well known in other secretory organs such as liver, kidney and intestine. The pancreas, representing a metabolically active organ with uptake and breakdown of essential nutritional components, changes its morphology and function with age. During childhood, the volume of the pancreas increases, reaching a plateau between 20 and 60 years, and declines thereafter. This decline involves the pancreatic parenchyma and is associated with decreased perfusion, fibrosis and atrophy. As a consequence of these changes, pancreatic exocrine function is impaired in healthy older individuals without any gastrointestinal disease. Five per cent of people older than 70 years and ten per cent older than 80 years have pancreatic exocrine insufficiency (PEI) with a faecal elastase-1 below 200 µg g-1 stool, and 5% have severe PEI with faecal elastase-1 below 100 µg g-1 stool. This may lead to maldigestion and malnutrition. Patients may have few symptoms, for example steatorrhoea, diarrhoea, abdominal pain and weight loss. Malnutrition consists of deficits of fat-soluble vitamins and is affecting both patients with PEI and the elderly. Secondary consequences may include decreased bone mineral density and results from impaired absorption of fat-soluble vitamin D due to impaired pancreatic exocrine function. The unanswered question is whether this age-related decrease in pancreatic function warrants therapy. Therapeutic intervention, which may consist of supplementation of pancreatic enzymes and/or vitamins in aged individuals with proven exocrine pancreas insufficiency, could contribute to healthy ageing.

Trends in indications, complications and outcomes for venous resection during pancreatoduodenectomy.

BACKGROUND: Pancreatoduodenectomy with superior mesenteric-portal vein resection has become a common procedure in pancreatic surgery. The aim of this study was to compare standard pancreatoduodenectomy with pancreatoduodenectomy plus venous resection at a high-volume centre, and to examine trends in management and outcome over a decade for the latter procedure. METHODS: This retrospective observational study included all patients undergoing pancreatoduodenectomy with or without venous resection at Oslo University Hospital between January 2006 and December 2015. Trends were evaluated by assessing preoperative clinical and radiological characteristics, as well as perioperative outcomes in three time intervals (early, intermediate and late). RESULTS: A total of 784 patients had a pancreatoduodenectomy, of whom 127 (16·2 per cent) underwent venous resection. Venous resection resulted in a longer operating time (median 422 versus 312 min; P = 0·001) and greater estimated blood loss (EBL) (median 700 versus 500 ml; P = 0·004) than standard pancreatoduodenectomy. The rate of severe complications was significantly higher for pancreatoduodenectomy with venous resection (37·0 versus 26·3 per cent; P = 0·014). The overall burden of complications, evaluated using the Comprehensive Complication Index (CCI), did not differ (median score 8·7 versus 8·7; P = 0·175). Trends in venous resection over time showed a significant reduction in EBL (median 1050 versus 375 ml; P = 0·001) and duration of hospital stay (median 14 versus 9 days; P = 0·011) between the early and late periods. However, despite an improvement in the intermediate period, severe complication rates returned to baseline in the late period (18 of 43 versus 9 of 42 versus 20 of 42 patients in early, intermediate and late periods respectively; P = 0·032), as did CCI scores (median 20·9 versus 0 versus 20·9; P = 0·041). CONCLUSION: Despite an initial improvement in severe complications for venous resection during pancreatoduodenectomy, this was not maintained over time. Every fourth patient with venous resection needed relaparotomy, most frequently for bleeding.

Impact of intrapancreatic or extrapancreatic bile duct involvement on survival following pancreatoduodenectomy for common bile duct cancer.

BACKGROUND: The clinicopathological factors that influence survival following pancreatoduodenectomy (PD) for common bile duct (CBD) cancer are not well known. This study aimed to investigate the effect of tumour involvement of the intrapancreatic versus extrapancreatic CBD on margin status, overall (OS) and disease-free (DFS) survival. METHODS: This was a retrospective study of patients who underwent PD for CBD cancer between 2001 and 2009. Pathological examination was performed according to a previously described standardized protocol based on axial slicing. Clinicopathological data and outcome in terms of margin status, DFS and OS were compared between cancers involving exclusively the intrapancreatic CBD (CBDin) and those involving the extrapancreatic CBD, in isolation or combined with invasion of the intrapancreatic part of the duct (CBDex). RESULTS: A total of 66 patients were enrolled. Most CBD cancers were locally advanced (97 per cent pathological (p) T3, 76 per cent pN1). Microscopic margin involvement (R1) was more frequent in CBDex than in CBDin cancers (34 of 39 versus 13 of 27; P = 0.001), more often multifocal (P < 0.001) and more frequently affected the periductal margin (P = 0.005). Venous resection was more often required for CBDex cancers (P = 0.009). CBDex cancers were associated with worse OS (median 21 versus 28 months; P = 0.020) and DFS (14 versus 31 months; P = 0.015), but the rate and site of recurrence did not differ. Metastasis to more than two lymph nodes was an independent predictor of OS and DFS. CONCLUSION: CBDex cancer is associated with a higher rate of R1 resection and venous resection after PD, and has a worse outcome than CBDin cancer.

Resection margins in pancreatic cancer.

The R1 rate and prognostic significance of microscopic margin involvement differ consistently between published series. This divergence results from a lack of consensus regarding various aspects of margin status assessment. Central to the controversies is the lack of clarity about what 'R1' exactly stands for. The current UICC definition--residual microscopic tumor--is possibly too general and invites divergent interpretations. Adherence to different diagnostic criteria for microscopic margin involvement and divergent terminology for the various margins of pancreatoduodenectomy specimens add to the confusion. Furthermore, recent studies demonstrated that the dissection technique and extent of tissue sampling influence the accuracy of margin assessment. Axial specimen slicing, extensive tissue sampling, and multicolored margin inking result in a significantly higher, more accurate R1 rate than when using traditional grossing techniques. Only when international consensus on these various aspects is reached will pathology data on margin involvement be reliable and can multicenter clinical trials produce compelling evidence that allows improved pancreatic cancer treatment.

Resection margin involvement and tumour origin in pancreatic head cancer.

BACKGROUND: Assessment of the origin of adenocarcinoma in pancreatoduodenectomy specimens (pancreatic, ampullary or biliary) and resection margin status is not performed in a consistent manner in different centres. The aim of this review was to identify the impact of such variations on patient outcome. METHODS: A systematic literature search for articles on pancreatic, ampullary, distal bile duct and periampullary cancer was performed, with special attention to data on resection margin status, pathological examination and outcome. RESULTS: The frequent reclassification of tumour origin following slide review, and the wide variation in published incidence of pancreatic (33-89 per cent), ampullary (5-42 per cent) and distal bile duct (5-38 per cent) cancers indicate that the histopathological distinction between the three cancer groups is less accurate than generally believed. Recent studies have shown that the wide range of rates of microscopic margin involvement (R1) in pancreatoduodenectomy specimens (18-85, 0-27 and 0-72 per cent respectively for pancreatic, ampullary and distal bile duct cancers) is mainly caused by differences in pathological assessment rather than surgical practice and patient selection. As a consequence of the existing inconsistency in reporting of these data items, the clinical significance of microscopic margin involvement in each of the three cancer groups remains unclear. CONCLUSION: Inaccurate and inconsistent distinction between pancreatic, ampullary and distal bile duct cancer, combined with inaccuracies in resection margin assessment, results in obfuscation of key clinicopathological data. Specimen dissection technique plays a key role in the quality of the assessment of both tumour origin and margin status. Unless the pathological examination is meticulous and standardized, comparison of results between centres and observations in multicentre trials will remain of limited value.

Resection margins and R1 rates in pancreatic cancer--are we there yet?

The prognosis of pancreatic cancer is poor, even for those patients who undergo surgical resection. The rate of local recurrence is high, despite the fact that in most series complete ('R0') resection is reported to be achieved in the majority of patients. The discrepancy between pathological assessment and clinical outcome indicates that microscopic margin involvement (R1) is frequently underreported, and potential causes for this are discussed in this review. Special emphasis is given to the variation that exists between currently used dissection techniques and their impact on the assessment of the resection margins in pancreatoduodenectomy specimens.

The impact of spontaneous tumour perforation on outcome following colon cancer surgery.

OBJECTIVE: The impact of spontaneous tumour perforation on survival following surgery for colon cancer is unclear. This study compares survival outcomes for patients with perforated colonic cancer with stage-matched nonperforated cancer. METHOD: A prospective histological database was searched for all patients undergoing resection for adenocarcinoma of the colon between 1996 and 2002. Patients with T4 cancer were selected and classified into those with spontaneous perforation at the tumour site and those with nonperforated tumour. Patients with synchronous colonic and rectal cancers, familial polyposis, inflammatory bowel disease, iatrogenic or remote colonic perforation were excluded. Histological variables were combined with clinical data obtained by case note review. Data were analysed for differences in demographics, histological variables, operative mortality, disease-free and overall survival. Multivariate analysis of factors predictive of overall survival in both groups was performed. RESULTS: Of 960 patients identified, 52 patients had spontaneous tumour perforation and 82 patients served as the T-stage matched control group. Overall survival at 2 years was 47% and 54% and at 5 years was 28% and 33% for perforated and nonperforated cancers respectively. Patients with perforated cancers were more likely to present with metastatic disease and undergo emergency surgery with a higher 30-day mortality. There was a trend towards reduced overall survival in the perforated group (P = 0.06), but no difference in disease-free survival (P = 0.43). On multivariate testing, 'emergency surgery' and 'age >75 years' were the only independent predictors of mortality in the perforated and nonperforated group respectively. CONCLUSION: Both perforated and nonperforated T4 colon cancers have a poor prognosis. Spontaneous perforation of the cancer is associated with reduced overall survival, due to higher 30-day mortality, but in itself does not appear to significantly impact on disease-free survival. Rather, it is the advanced oncological stage at which perforated cancers present that determines outcome.

Pancreatic polypeptide cell hyperplasia of the pancreas.

A case of pancreatic polypeptide cell hyperplasia in a 76-year-old man who presented with subacute bowel pseudo-obstruction is reported. A computed tomography scan incidentally showed a pancreatic head lesion that was resected by pancreaticoduodenectomy. Histological examination showed expansion of the endocrine pancreas with increased numbers of pancreatic polypeptide cells in irregularly enlarged islets, ragged endocrine cell clusters, ductulo-insular complexes and microadenomas. The clinicopathological features of this rare and poorly understood condition are discussed.

Fas ligand expression in Hodgkin lymphoma.

Because previous investigations suggested involvement of the Fas ligand (FasL) in the selection process in the follicular dendritic cell (FDC)-associated cell cluster of the germinal center, we investigated the expression of FasL in Hodgkin lymphoma (HL) on protein and RNA level, while considering the Epstein-Barr virus status of the Hodgkin and Reed-Sternberg (HRS) cells. Tumor tissue from 47 patients with classic HL (32 nodular sclerosis [NS], 11 mixed cellularity [MC], and 4 lymphocyte-rich [LR]) was analyzed by immunohistochemistry for FasL, Fas, CD21, and CD23 and by Western blotting for FasL. FasL mRNA was detected by an exon 4-specific oligonucleotide and Epstein-Barr virus infection by in situ hybridization for Epstein-Barr virus early RNAs (EBER). Western blotting showed soluble and membrane-bound forms of FasL. Immunohistochemistry showed FasL expression in virtually all HRS of 94% of NS cases and 82% of MC cases. FasL expression did not correlate with the Epstein-Barr virus status of the HRS. Low FasL protein expression was found in some HRS of LR cases. FasL mRNA was detected in 39% of NS, 46% of MC, and 33% of LR cases. Seventy percent to 90% of the HRS cells expressed Fas. CD21 immunohistochemistry showed disrupted FDC networks in the tumor tissue with reduced and virtually absent expression of CD23 and FasL. These observations suggest that FasL expression in HRS cells and the absence of FasL in the FDC cluster represent a disturbed microenvironment that may be involved in the pathogenesis of HL.

The roles of cell surface attachment molecules and coagulation Factor X in adenovirus 5-mediated gene transfer in pancreatic cancer cells.

Transduction of 11 pancreatic cancer cell lines with a replication-deficient adenovirus 5 expressing enhanced green fluorescent protein (Ad5EGFP) was analyzed and variable EGFP levels were observed, ranging from <1% to ∼40% of cells transduced, depending on the cell line. Efficient Ad5EGFP transduction was associated mainly with higher levels of cell surface Coxsackie and adenovirus receptor (CAR) but not with expression of α(v)ß(3) and α(v)ß(5) integrins and was fiber dependent. Reduction of CAR by RNA interference resulted in a corresponding decrease in Ad5EGFP transduction. Pre-treatment of Ad5EGFP with blood coagulation Factor X increased virus entry even in the presence of low CAR levels generated by RNA interference, suggesting a potential alternative route of Ad5 entry into pancreatic cancer cells. Immunohistochemistry carried out on 188 pancreatic ductal adenocarcinomas and 68 matched controls showed that CAR was absent in 102 (54%) of adenocarcinomas, whereas moderate and strong staining was observed in 58 (31%) and 28 (15%) cases, respectively. Weak or absent CAR immunolabeling correlated with poor histological differentiation of pancreatic cancer. In normal tissue, strong immunolabeling was detected in islet cells and in the majority of inter- and intralobular pancreatic ducts.

Evolution of idiopathic fibrosing pancreatitis--MRI features.

Here, the clinical and imaging features of idiopathic fibrosing pancreatitis are described, including a description of the evolution of MRI features in a patient treated successfully with biliary stenting alone. Thus, not all masses of the pancreatic head in the paediatric population need to be managed surgically.

Predictors of malignant potential of cystic lesions of the pancreas.

BACKGROUND: Cystic lesions of the pancreas (CLP) are a diagnostic dilemma, the correct characterisation of which determines surgical management. METHODS: From 1995 to 2005, radiology and pathology records were reviewed for the presence of CLP. CLP were divided into three groups; Group 1: Benign, Group 2: Pre-malignant, and Group 3: Malignant. RESULTS: Seventy-nine of 121 patients were included [Group 1: n=46, Group 2: n=10, Group 3: n=23], with a median age at diagnosis of 68 (31-92) years. The median follow-up period was 24 (14-84) months. On univariate analysis, female gender (p=0.04), jaundice (p<0.01), raised serum ALT concentration (p=0.03), cyst size (> or = 2.5 cm) (p<0.01), and biliary duct dilatation (p<0.01) were associated with malignant potential. Benign cysts were more likely to present incidentally (p<0.01). On multi-variate analysis, cyst size (> or =2.5 cm) was an independent predictor of malignant potential. Sub-group analysis revealed that cysts <2.5 cm in the head of the pancreas with evidence of biliary obstruction (either abnormal liver function; raised ALT [p=0.01], ALP [p=0.01], total bilirubin [p=0.02], and/or biliary duct dilatation [p<0.01]) were associated with malignant potential. CONCLUSION: Cyst size > or =2.5 cm on computer tomography imaging was an independent predictor of pre-malignant and malignant pancreatic cysts. Cyst size and the presence of biliary obstruction predict potentially malignant cysts of the head of the pancreas, which require surgical management.

Redefining the R1 resection in pancreatic cancer.

BACKGROUND: Resection margin (RM) status in pancreatic head adenocarcinoma is assessed histologically, but pathological examination is not standardized. The aim of this study was to assess the influence of standardized pathological examination on the reporting of RM status. METHODS: A standardized protocol (SP) for pancreaticoduodenectomy specimen examination, involving multicolour margin staining, axial slicing and extensive tissue sampling, was developed. R1 resection was defined as tumour within 1 mm of the RM. A prospective series reported according to this protocol (SP series, n = 54) was compared with a historical matched series in which a non-standardized protocol was used (NSP series, n = 48). RESULTS: Implementation of the SP resulted in a higher R1 rate overall, and for pancreatic (22 of 26 85 per cent) compared with ampullary (four of 15) and bile duct (six of 13) cancer. Sampling of the circumferential RM was more extensive in the SP series and correlated with RM status. RM involvement was often multifocal (14 of 32), affecting the posterior RM most frequently (21 of 32). Survival correlated with RM status for the entire SP series (P < 0.001), but not for the NSP series. There was a trend towards better median and actuarial 5-year survival after R0 resection in the SP pancreatic cancer subgroup. CONCLUSION: Standardized examination influences the reporting of RM status.

Cyclooxygenase-2 expression associated with severity of PanIN lesions: a possible link between chronic pancreatitis and pancreatic cancer.

BACKGROUND: Cyclooxygenase-2 (COX-2) is a key modulatory molecule in inflammation and neoplasia. Increasing evidence suggests a role for COX-2 in pancreatic cancer (PAC). However, expression of COX-2 in pancreatic intraepithelial neoplasia (PanIN), the precursor lesion of PAC which is often present in chronic pancreatitis (CP), has received little attention. METHOD: COX-2 immunostaining was performed on sections of PAC (n = 26), CP (n = 34), PanIN (n = 68) and normal pancreas (n = 11). Sections were also stained for macrophages (CD68), activated pancreatic stellate cells (alphaSMA), and collagen (Sirius Red) as markers of fibrosis. Semiquantitative scoring was based on the extent and intensity of immunostaining. RESULTS: COX-2 expression was increased in PAC compared to normal (p = 0.02) with 89% of cases exceeding COX-2 immunostaining in normal ducts. In PanIN lesions, COX-2 expression increased with escalating severity of the PanIN change (p < or = 0.01). COX-2 expression was increased in PanIN-2/3 compared to normal pancreas and CP (p < or = 0.001). In ducts of CP, COX-2 expression did not differ from that in normal tissue. There was no association between COX-2 expression and clinicopathological variables. CONCLUSION: The high level of COX-2 expression in PanIN lesions suggests that this enzyme could be a therapeutic target at a non-invasive stage of pancreatic carcinogenesis and feasible for chemoprevention in CP.

Sideroelastosis pulmonum: cause or consequence of the pulmonary veno-occlusive disease?

In a 66-year-old patient morphologic features of a pulmonary veno-occlusive disease were found. Besides, a striking foreign body giant cell reaction with phagocytosis of altered elastic tissue in totally or partially occluded venous blood vessels resembled the so-called pulmonary sideroelastosis, an entity originally described by Ceelen. In this case report, a possible pathogenetic relation between sideroelastosis pulmonum and pulmonary veno-occlusive disease is discussed.

[Computerized tomography-guided fine needle biopsy. Current perspectives in biopsy diagnosis]. / Computertomographisch gesteuerte Feinnadelbiopsie. Neue Dimensionen der bioptischen Diagnostik.

253 cutting needle biopsies from 240 patients (151 men, 89 women; mean age 68 [23-84] years) were reviewed. The biopsies were taken under computed (CT) tomography guidance from various parts of the body (pancreas: n = 48, liver: n = 32, other upper abdominal organs: n = 5, mediastinum: n = 27, lung: n = 37, kidney: n = 13, adrenals: n = 8, retroperitoneum: n = 35, pelvis: n = 37, pleura: n = 4, chest wall: n = 6). The data were evaluated retrospectively. In 86% of cases meaningful clinical information was obtained in the form of a definite diagnosis (74%) or of a presumptive or differential diagnosis (12%). As a rule it was possible to assess the degree of malignancy and to ascertain the exact nature of the condition; the latter was true not only of malignant but also of benign lesions. In the majority of cases this had important implications. In just under 14% of all cases the target area was missed and no diagnosis could be made. To raise the chances of obtaining a representative biopsy multiple punctures were performed in 44% of the CT-guided biopsies. In 61 cases (24%) the diagnosis was checked against further tissue samples and in four cases the diagnosis based on needle biopsy material had to be revised. In all, only two complications were recorded; both were mild and did not require treatment. In terms of diagnostic value, CT-guided fine needle biopsies proved considerably superior to conventional cytological studies of fine needle aspiration material.

Fas ligand expression in the germinal centre.

Whereas the importance of the Fas/FasL system in the regulation of T-cell homeostasis is well established, it is not yet clear if FasL is involved in B-cell regulation, especially in the clonal selection of B lymphocytes in the germinal centre (GC). This study therefore investigated the expression of FasL protein in tonsils and lymph nodes with lymphofollicular hyperplasia by western blotting and immunohistochemistry. In all the samples examined, western blot analysis showed FasL proteins of 33 and 52 kD, which presumably correspond to membrane-bound and soluble forms of the FasL protein. Immunohistochemically, FasL was found in a limited number of cells confined to a cluster in the light zone of the GC. The signal showed a delicate meshwork-like pattern of branching processes enmeshing the centrocytes and the few centroblasts of the light zone. In serial sections, the immunostaining pattern for FasL was found largely to coincide with the CD23 staining of follicular dendritic cells (FDCs), which are typically located in the light zone. In contrast, the FasL signal did not correspond to the distribution of the CD4-positive GC T-cells. In conclusion, expression of FasL in lymphofollicular hyperplasia seems to be largely confined to the light zone of the GCs, where selection of FDC-associated centrocytes is known to occur. These observations thus suggest that FasL is involved in selection processes of the B-cell system.

Angiogenesis and cathepsin expression are prognostic factors in pancreatic adenocarcinoma after curative resection.

BACKGROUND: Curative resection of pancreatic adenocarcinoma is the only clinical parameter related to a favorable prognosis while other clinicopathological parameters fail. To evaluate whether angiogenesis, vascular endothelial growth factor (VEGF) or certain tumor proteases, e.g., cathepsin B (CTSB) and L (CTSL), are factors of prognostic relevance, we investigated their expression in patients with long- and short-term survival after curative resection (RO) because of pancreatic adenocarcinoma. METHODS: Twenty-nine tissue samples from patients with adenocarcinoma of the pancreas were examined. The patients were selected in a long-term survival group with a survival > or = 24 mo (n = 18) and a shortterm survival group of patients, who died within 8 mo after surgery because of their malignancy (n = 11). The microvessel quantification was performed immunohistochemically using a monoclonal anti-CD34 antibody. VEGF, CTSB, and CTSL expressions was studied using polyclonal antibodies (PAbs). RESULTS: The median microvessel density (MVD) was 75 (range 39-182). MVD correlated significantly with the survival time after surgery (p = 0.0132) but not with clinicopathological parameters. In cancer cells, VEGF was positive in 82.8% and showed significant correlation with the MVD (p = 0.0002) and survival time (p = 0.0395). Positive immunoreactivity could be obtained for 96.5% for CTSB and 84.2% for CTSL. Expression of both proteases correlated significantly with the survival time after surgery (CTSB p = 0.0002, CTSL p = 0.0001). Furthermore, CTSB expression correlated with invasion of the perineural space. Thus, a short postoperative survival correlated with a high MVD, and highly expressed VEGF, CTSB, and CTSL. No significant correlation between MVD, VEGF, as well as CTSL and clinicopathological parameters was found. For routinely assessed markers (e.g., TNM-stage, UICC-stage, and so on) no significant correlation with survival time was found in this small group of patients. CONCLUSION: These findings indicate that the MVD, VEGF, CTSB, and CTSL are prognostic factors after curative resection, whereas other parameters (TNM, UICC, and so on) failed to show prognostic relevance in our group of patients. Furthermore, the correlation between MVD and VEGF underlines the importance of this growth factor for angiogenesis and tumor growth. The correlation between CTSB and perineural invasion demonstrates the involvement of cathepsins in local tumor invasion.