Page 151 - Read Online

Page 218 Grewal et al. Art Int Surg 2023;3:217-32 https://dx.doi.org/10.20517/ais.2023.28

intelligence algorithms, radiomics represents a promising biomarker avenue for promoting enhanced and tailored
management of HPB cancers, with the potential to improve long-term outcomes for patients.

Keywords: Pancreatic neoplasms, pancreatic ductal adenocarcinoma, hepatocellular carcinoma,
cholangiocarcinoma, radiomics, artificial intelligence

INTRODUCTION
Hepatobiliary and pancreatic malignancies comprise a heterogeneous group of diseases that rank amongst
[1]
the leading causes of cancer-related deaths worldwide . Despite improvements in cancer surveillance,
imaging, and treatment, the prognosis of these patients remains poor, with the 5-year survival rate of
pancreatic cancer [pancreatic ductal adenocarcinoma (PDAC)] and biliary tract carcinomas (BTCs)
reported of only 10% and 5%-18%, respectively . The prognosis of localized hepatocellular carcinoma
[2,3]
[4]
(HCC) is marginally better at 33%; however, it drops off to 2% in the context of metastatic disease .
Poor outcomes are predominantly driven by a delay in the diagnosis of these diseases . Diagnostic
[1]
challenges of hepatobiliary and pancreatic (HPB) malignancies are multifaceted. Firstly, due to the
asymptomatic nature of the disease, these patients are diagnosed with more advanced diseases, with only
43% of hepatic cancer and 20% of pancreatic cancers being diagnosed at an early stage . Secondly, there is
[5,6]
a lack of screening modalities (PDAC) or the current screening strategies are not very accurate (HCC), e.g.,
the ultrasound-based protocol for HCC has a sensitivity of 47%-63% for detection of early-stage disease .
[7]
Biomarkers for screening of these diseases are lacking . Additionally, pancreatic cancer often mimics
[8,9]
benign lesions on imaging, which can result in misdiagnosis and delay in the delivery of appropriate care .
[6]
It is estimated that nearly 5%-11% of all patients undergoing pancreaticoduodenectomies for pancreatic
cancer turn out to have benign lesions . Altogether, there is a critical need for noninvasive biomarkers to
[10]
facilitate earlier diagnosis of hepatobiliary and pancreatic malignancies.

Once a diagnosis is established, tumor characterization (tumor grade, presence of nodal disease, extent of
local invasion, and molecular profile) is essential in determining appropriate care. This is vital for
sequencing of systemic therapy and surgical planning [11,12] . While helpful, current imaging modalities are
limited by their accuracy in determining these features. Furthermore, the selection of appropriate systemic
therapies presents a challenge in the management of these diseases. Currently, clinical tools to predict
treatment response are absent, and it is not determined until the patient undergoes resection and a
histopathological examination of the specimen is performed. If the disease was resistant to the administered
chemotherapeutics, unfortunately, the patient gained no benefit from this therapy, while allowing time for
resistant clones to proliferate and result in the progression of disease. For instance, in the setting of
metastatic PDAC, Nab-paclitaxel combined with gemcitabine represents the standard of care; however,
there is considerable heterogeneity among patients with respect to duration of treatment (0.1-21.9 months),
secondary to the significant toxicities of these therapies, as demonstrated by the MPACT trial, as well as
treatment response, with the SIEGE trial reporting almost a fifth of patients failing to reach their first
treatment response assessment [13,14] . As such, there is a critical need to robustly validate biomarkers to both
spare patients from toxicities associated with treatments that they are unlikely to respond to and to identify
new avenues for targeted treatments.

Recently, multiple biomarkers have shown promise in hepatobiliary and pancreatic malignancies, including
circulating tumor cells, circulating tumor DNA, proteins, and radiomics [9,15-17] . The identification and
validation of biomarkers could aid in patient stratification, treatment planning, and prediction of response