Survival Outcome and Prognostic Factors for Pancreatic Acinar Cell Carcinoma: Retrospective Analysis from the German Cancer Registry Group

Simple Summary Less than 1% of all pancreatic malignancies are acinar cell carcinomas. Based on data from the German Cancer Registry Group, we performed a comparative analysis of characteristics and prognostic factors of pancreatic acinar cell carcinoma and the most common type of pancreatic cancer—pancreatic ductal adenocarcinoma. Compared to pancreatic ductal adenocarcinoma, patients with pancreatic acinar cell carcinoma were younger at the time of diagnosis and the percentage of males was higher. The prognosis of patients with pancreatic acinar carcinoma was better than that of patients with pancreatic ductal adenocarcinoma. Surgical resection was the strongest positive prognostic factor for pancreatic acinar cell carcinoma. The study shows that pancreatic acinar cell carcinoma has features distinct from pancreatic ductal adenocarcinoma. Radical resection should be advocated, whenever feasible. Abstract Background: Pancreatic acinar cell carcinoma (PACC) is a distinct type of pancreatic cancer with low prevalence. We aimed to analyze prognostic factors and survival outcome for PACC in comparison to pancreatic ductal adenocarcinoma (PDAC), based on data from the German Cancer Registry Group. Methods: Patients with PACC and PDAC were extracted from pooled data of the German clinical cancer registries (years 2000 to 2019). The distribution of demographic parameters, tumor stage and therapy modes were compared between PACC and PDAC. The Kaplan–Meier method and Cox regression analysis were used to delineate prognostic factors for PACC. Propensity score matching was used to compare survival between PACC and PDAC. Results: There were 233 (0.44%) patients with PACC out of 52,518 patients with pancreatic malignancy. Compared to PDAC, patients with PACC were younger (median age 66 versus 70, respectively, p < 0.001) and the percentage of males was higher (66.1% versus 53.3%, respectively, p < 0.001). More patients were resected with PACC than with PDAC (56.2% versus 38.9%, respectively, p < 0.001). The estimated overall median survival in PACC was 22 months (95% confidence interval 15 to 27), compared to 12 months (95% confidence interval 10 to 13) in the matched PDAC cohort (p < 0.001). Surgical resection was the strongest positive prognostic factor for PACC after adjusting for sex, age, and distant metastases (hazard ratio 0.34, 95% confidence interval 0.22 to 0.51, p < 0.001). There was no survival benefit for adjuvant therapy in PACC. Conclusions: PACC has overall better prognosis than PDAC. Surgical resection is the best therapeutic strategy for PACC and should be advocated even in advanced tumor stages.


Introduction
Pancreatic acinar cell carcinoma (PACC) is a distinct type of exocrine pancreatic cancer that arises from the acinic cells. It presents microscopically as a solid, cellular neoplasm with minimal stroma [1], (Figure 1). The prevalence of PACC has been estimated to be below 1% of all pancreatic malignancies [2]. Compared to patients with the most common pancreatic cancer, pancreatic ductal adenocarcinoma (PDAC), patients with PACC are on average younger and have a better overall prognosis [2,3]. Clinical evidence on PACC is based on case reports and small institutional series as well as a few registry-based studies that analyzed larger cohorts of patients with PACC [2][3][4][5][6][7]. Many of these studies point out surgical resection as the most effective therapeutic strategy [2][3][4][5]7]. The survival benefit of systemic therapy is more controversial [3,8,9].

Introduction
Pancreatic acinar cell carcinoma (PACC) is a distinct type of exocrine pancreatic cancer that arises from the acinic cells. It presents microscopically as a solid, cellular neoplasm with minimal stroma [1], (Figure 1). The prevalence of PACC has been estimated to be below 1% of all pancreatic malignancies [2]. Compared to patients with the most common pancreatic cancer, pancreatic ductal adenocarcinoma (PDAC), patients with PACC are on average younger and have a better overall prognosis [2,3]. Clinical evidence on PACC is based on case reports and small institutional series as well as a few registry-based studies that analyzed larger cohorts of patients with PACC [2][3][4][5][6][7]. Many of these studies point out surgical resection as the most effective therapeutic strategy [2][3][4][5]7]. The survival benefit of systemic therapy is more controversial [3,8,9]. Because of the low prevalence of PACC, clinical practices well studied in PDAC are often applied also to PACC. However, PACC has distinct clinical, histologic, and molecular features that make a differential approach necessary [1,2,4,10]. Especially relevant are the questions of long-term survival after resection of locally advanced tumors, resection of synchronous or metachronous distant metastases, as well as the benefit of different perioperative and palliative therapies. To expand our knowledge on a rare malignancy such as PACC, registry-based studies are the most feasible.
In this study, we aimed to analyze prognostic factors and survival outcomes for PACC based on data from the German clinical cancer registries. To delineate the specifics of PACC, a comparative analysis to PDAC was performed.

Study Design and Ethics
This is a retrospective, registry-based study. It was approved by the local ethics committee of the University of Lübeck, Germany (Reference Number: 20-319). Because of the low prevalence of PACC, clinical practices well studied in PDAC are often applied also to PACC. However, PACC has distinct clinical, histologic, and molecular features that make a differential approach necessary [1,2,4,10]. Especially relevant are the questions of long-term survival after resection of locally advanced tumors, resection of synchronous or metachronous distant metastases, as well as the benefit of different perioperative and palliative therapies. To expand our knowledge on a rare malignancy such as PACC, registry-based studies are the most feasible.

German Clinical Cancer Registries
In this study, we aimed to analyze prognostic factors and survival outcomes for PACC based on data from the German clinical cancer registries. To delineate the specifics of PACC, a comparative analysis to PDAC was performed.

Study Design and Ethics
This is a retrospective, registry-based study. It was approved by the local ethics committee of the University of Lübeck, Germany (Reference Number: 20-319).

German Clinical Cancer Registries
There is a longstanding tradition of hospital-based clinical cancer registration in Germany. Since 2013, clinical cancer registries have been established for quality assurance purposes in all federal states of Germany as part of the German National Cancer Plan. Therefore, in most of the federal states of Germany the preexisting epidemiological cancer registries have been expanded into population-based clinical registries [11]. The Society of German Tumor Centers (ADT) is the head organization that coordinates the German Cancer Registry Group on a national level. There is a uniform minimal data set for all oncological entities [12]. This includes items on demographics (age, sex), tumor entity, topography, and histology according to the International Classification of Diseases for Oncology (ICD-O) [13], TNM classification [14], therapy (operation, systemic therapy, radiation), and survival follow-up. The combined data from 17 clinical registries from the years 2000 to 2019 were used according to the data use rules of the German Cancer Registry Group.

Statistical Methods
Differences in the overall distribution of the various parameters between PACC and PDAC were delineated. Additionally, multivariable analysis with covariables age, sex, distant metastases, and operation was performed. With the subgroup of patients with upfront surgery (resection without prior neoadjuvant therapy), univariable survival analysis with covariables age, sex, T-stage, lymph node metastases, distant metastases, vessel invasion, lymph vessel invasion, resection status, adjuvant therapy was performed. Data processing and statistical analysis were performed with R version 3.3.3 [16]. Descriptive statistics with median and interquartile range of continuous variables and absolute number and percentage of total for categorical variables were used. Univariable logistic regression was applied in the comparison of the distribution of the parameters studied between PACC and PDAC. The propensity score was calculated via logistic regression with covariables age, sex, therapy, T-stage, lymph node metastases, distant metastases, resection status, grading, year of diagnosis, survival status. Propensity score nearest neighbor matching was performed. The R package MatchIt was used [17].

Survival Analysis
For patients with available survival data, propensity score matching of PACC to PDAC in ratio 1:3 was performed. Survival analysis for PACC and PDAC was performed with the matched data. Univariable survival analysis for age, sex, distant metastases, grading, and therapy type was applied to the whole data set. Kaplan-Meier method and log-rank test were applied in the univariable survival analysis. Additionally, multivariable Cox regression with covariables age, sex, distant metastases, and operation was performed. With the subgroup of patients with upfront surgery (resection without prior neoadjuvant therapy), univariable survival analysis with covariables age, sex, T-stage, lymph node metastases, distant metastases, vessel invasion, lymph vessel invasion, resection status, adjuvant therapy was performed. Follow-up was estimated with the reversed Kaplan-Meier method. Significance level 0.05 and confidence interval 95% were set.

Comparison PACC and PDAC
PACC patients were compared to a cohort of 37,940 patients with PDAC (72.2% of all 52,518 patients). The median age at time of diagnosis of PDAC patients was 70 years compared to 66 years of PACC patients (p < 0.001). The proportion of male patients was higher in the PACC cohort (66.1% versus 53.3%, p < 0.001). More patients were resected in the PACC cohort (56.2% versus 38.9%, p < 0.001). Of those resected, more patients received adjuvant chemotherapy in the PACC cohort (38.2% versus 30.1%, p = 0.045). PACC patients underwent distal pancreatectomy more often than PDAC patients (29 versus 8.2%, respectively, p < 0.001) ( Table 1). The estimated overall survival after diagnosis for PACC patients was 22 (15 to 27) months, and for all PDAC patients, this was 8.1 (8.07 to 8.13) months (p < 0.001).
The 218 PACC patients with available survival data were matched to 654 patients with PDAC. The estimated overall survival of the matched PDAC cohort was 12 (10 to 13) months ( Figure 2A). Females had better prognosis in PACC (31 versus 20 months, respectively, p = 0.01) compared to males but not in PDAC (13 versus 11 months, respectively, p = 0.163). There were several differences between PACC and the matched PDAC in the subgroup of patients with upfront surgery: there was no statistically significant difference in the estimated survival of resected patients without (M0) and with distant metastases (M1) in PACC (29 versus 32 months, respectively, p = 0.709) but significantly shorter survival for M1 in PDAC (24 versus 9 months, respectively, p < 0.001); no significant difference in resection margin status in PACC R0 versus R+ (34 versus 23 months, respectively, p = 0.208) but significant difference in PDAC (29 versus 12 months, respectively, p < 0.001); no significant difference with adjuvant therapy versus no adjuvant therapy in PACC (36 versus 34 months, respectively, p = 0.48) but survival benefit for PDAC patients with adjuvant therapy (30 versus 16 months, respectively, p < 0.001).

Discussion
The results of the current analysis are in line with the results of previous studies, showing that PACC is a rare type of pancreatic cancer, occurring more often in males and at a younger age, compared to PDAC. Estimated overall median survival was better in PACC compared to a matched PDAC cohort (22 versus 12 months, respectively, p < 0.001). Resection, younger age, and female sex were independent positive prognostic factors. No survival benefit could be shown for systemic therapy.
Most of the large-scale studies on PACC are based on data from the American National Cancer Database (NCDB) as well as from the Surveillance, Epidemiology, and End Results (SEER) registry [3,5,8,[18][19][20]. Some of these studies analyze overlapping cohorts, focusing on different aspects [5,8,[18][19][20]. Table 4   Demographic characteristics of the European cohort described in this study were comparable to those reported of cohorts from the United States. Most of the NCBI/SEERbased studies reported a median age of PACC patients about 65 years. Patients in the two Japanese publications, Kitagami et al. and Takahashi et al., were younger with median ages 60 and 61 years [9,22]. In the current study, the median age for PACC patients was 66 years, compared to 70 years for PDAC patients (p < 0.001). Similarly, a higher median age for PDAC was reported by Wisnoski [21].
All but one study reported the predominance of the male sex with 54-74%. In our study, the proportion of males in PACC was 66.1% and, thus, significantly higher than in PDAC (53.3%, p < 0.001). A significantly higher male proportion in PACC than in PDAC was reported also in Wisnoski et al. (53.6 vs. 48.4%, respectively, p < 0.01) and Schmidt et al. (63.5 vs. 49.9%, respectively, p < 0.001) [2,3].
The proportion of PACC patients with distant metastases was lower than that of PDAC patients (33.9% vs. 40%, respectively, p = 0.002). However, because of the high proportion of unknown metastases status Mx in both groups (20.6% in PACC versus 25.8% in PDAC), the actual distribution might be different. In resected patients, lymph node metastases N1 were more common in PDAC (63.7%) than in PACC (42.7%, p < 0.001). Here, the proportion of unknown lymph node status Nx was much lower (4.6% PACC and 7.8% PDAC) and, thus, would not impact the N0 to N1 distribution significantly. Schmidt  and lymph node metastases N1 for PACC, but again here the proportion of Mx and Nx was too high to draw a conclusion about the actual distribution [3].
PACC showed a much better prognosis with an estimated median survival time of 22 months (95% CI 15 to 27) compared to the PDAC cohort with 8.1 months (95% CI 8.07 to 8.13). To account for the differences in stage, sex, and age distribution as well as differences in therapies applied, we performed a propensity score matching of 1:3 PACC to PDAC patients. After matching, the difference in overall median survival was smaller but still significantly better in PACC (22 vs. 12 months, p < 0.001). Resection was the strongest positive prognostic factor for PACC, after adjustment for age, sex, and distant metastases (OR 0.34, 95% CI 0.22 to 0.51, p < 0.001).
Among resected PACC patients, lymph node metastases were a negative prognostic factor. Positive resection margin was not associated with survival, but the analysis might be biased and underpowered because of missing data. It is noteworthy that distant metastases were a negative prognostic factor in resected PDAC patients but not in PACC patients. Though this result should be interpreted with caution due to potential bias, metastases resection should be given consideration for PACC patients. Similar to our study, Landa et al. reported 12% of 566 resected PACC patients to be stage IV, with an estimated 5-year survival of 19% versus 4% (p < 0.001) of not resected stage IV patients [5]. There have been several case reports of successful curative resections of metastases in PACC [24][25][26][27][28][29]. Ohara et al.  [24]. Besides resection, radiofrequency ablation has been reported to be successful in treating liver metastases in PACC patients and should be given consideration in the multimodal treatment strategy [30,31].
Our study did not demonstrate any survival benefit of systemic therapy in PACC, while adjuvant as well as palliative systemic therapy was associated with better prognosis in the matched PDAC cohort. In a recent study, Patel et al. demonstrated a survival benefit for adjuvant chemotherapy in PACC in a multivariable Cox regression analysis [8].
Because of the small number of patients (N = 131) and differently distributed missing data among the variables, we did not perform a multivariable analysis for the subgroup of resected patients. Furthermore, the most common regime of adjuvant chemotherapy given for PACC was gemcitabine monotherapy in 30 patients (60% of patients with adjuvant chemotherapy). The results of several previous studies suggest that gemcitabine monotherapy might not lead to a response in PACC patients but, rather, platin-based regimes should be given preference [4,9,23,[32][33][34]. Thus, the lack of overall survival benefit of adjuvant therapy in our study might also be due to the high proportion of patients with gemcitabine monotherapy.
Studies of the molecular biology of PACC have shown frequent aberrant DNA methylation and chromosomal abnormalities, and it has been reported that about one third of PACC have targetable genetic alterations in genes such as BRAF, BRCA2, SMAD4, RAF1, NTRK1 [10,35]. Corresponding targeted therapies have been successfully applied [32,36,37]. Thus, molecular profiling should have an integral role in guiding therapeutic decisions in patients with PACC.

Conclusions
It is important to bear in mind that PACC has features distinct from PDAC. Radical surgical resection should be advocated, including resection of metastases, whenever feasible. Though the role of systemic therapy is not well defined, it should be considered at least for patients with advanced disease.