Contrast-enhanced ultrasound as a non-invasive diagnostic modality for pancreatic ductal adenocarcinoma: The question of Ki67 for study validation

Abstract

This editorial comments on Yang et al’s article that reported a correlation between dynamic contrast-enhanced ultrasound (CEUS) quantitative parameters and Ki67/tumor differentiation. The validation of CEUS as a diagnostic modality in this study deserves merit. However, it raises interesting points of discussion: (1) Since pancreatic cancer is an overarching term that includes conventional pancreatic ductal adenocarcinoma (PDAC), other subtypes, and neuroendocrine neoplasms (NENs), the inclusion/exclusion criteria require better clarification; (2) Most PDACs are grade 1-2 which contrasts with Yang et al’s study where 46% were grade 3; (3) Ki67 is officially recognized for grading NENs, but not for PDAC; (4) Hotspots are selected for the Ki67 grading of NENs. However, for other tumors (e.g., breast carcinoma), the average count or hotspots are used; (5) There is no agreement for defining high-grade Ki67 cut-off for non-NENs; reports range from 10% to 50%; and (6) Ki67 reflects cellular proliferation but is not always the most important indicator for biologic aggressiveness. That notwithstanding, since the ratification of Ki67 for prognosis in NENs was based on survival outcomes, the real gold standard should be survival, instead of using Ki67 as a surrogate gold standard. In conclusion, the validation of CEUS parameters for PDAC is a work in progress. CEUS is valuable in assessing PDAC but should be viewed as augmenting other modalities such as computed tomography, magnetic resonance imaging, positron emission tomography and endoscopic ultrasound.

Key Words: Contrast-enhanced ultrasound; Quantitative parameters; Imaging modalities; Pancreatic ductal adenocarcinoma; Ki67

Core Tip: Correlations have been reported between contrast-enhanced ultrasound (CEUS) quantitative parameters and Ki67/tumor differentiation. However, the validation process is still a work in progress. Meticulous attention should be paid to subtypes of pancreatic ductal adenocarcinoma selected for testing, the suitability of Ki67 as the gold standard, the counting methodology and grade cut-offs. Overall, CEUS appears valuable in assessing pancreatic ductal adenocarcinoma. However, it should be regarded one more additional tool along with other imaging modalities such as computed tomography, magnetic resonance imaging and positron emission tomography scan. CEUS can also be augmented by endoscopic ultrasound-guided fine needle biopsy for tissue procurement. In certain situations, detective flow imaging endoscopic ultrasonography could be an alternative to CEUS.

INTRODUCTION

Dynamic contrast-enhanced ultrasound (CEUS) offers quantitative perfusion data for the diagnosis and monitoring of various diseases[1,2]. It has found clinical application for both neoplastic and non-neoplastic conditions in virtually every body system[1-3]. From an oncologic perspective, it can provide insights into tumor vascularization and thus help in assessing mass characteristics. This editorial comments on the article by Yang et al[4] who investigated the correlation between quantitative parameters of CEUS and Ki67/tumor differentiation in an attempt to validate CEUS as a non-invasive diagnostic modality for pancreatic ductal adenocarcinoma (PDAC). The goal was to not only use CEUS for circumventing the risks associated with invasive biopsy procedures but also provide a more accurate basis for treatment decisions and prognosis assessment, thereby improving outcomes and quality of life.

To better define its place in the diagnostic armamentarium, CEUS has been compared to other imaging modalities such as computed tomography (CT), magnetic resonance imaging (MRI), and positron emission tomography (PET)[1,2,5-8]. From a historical viewpoint, the first reported medical use of echo contrast dates back to the 1960s when injection of saline was utilized to study heart valves[1,5]. Since then, various solutions have been utilized ranging from hand-agitated bubble suspensions to the current commercial ultrasound contrast agents that use bubbles of standardized size. Over time, the applications of the technique extended beyond cardiovascular disease to other organs including both neoplastic and non-neoplastic diseases[1,2,5-8]. However, with increasing utilization comes the requirement for ratification. It is therefore pertinent to conduct validation assessments for the expanding role of CEUS in medical imaging, which in this context is with particular reference to PDAC.

DISCUSSION

CEUS

Ultrasound imaging is a common imaging technology in the world rivaling CT and MRI due largely to its favorable characteristics e.g., easy performance, real-time scanning, non-radiating energy, and cost-effectiveness[9,10]. With respect to the pancreas, endoscopic ultrasound (EUS) may have an advantage over transabdominal ultrasound since it can overcome problems caused by intervening gas, mural bone and fat. It can therefore produce better spatial resolution, a very useful feature for small lesions. However, most lesions appear hypoechoic on EUS which limits the differential diagnosis. EUS is further limited by the scope of conventional color-Doppler and power-Doppler imaging. These conventional EUS Doppler modes are effective at assessing large vessels with fast-flowing blood but are not good for visualizing fine vessels and slow flow at the tissue perfusion level[11]. The advent of CEUS and its subsequent technical improvements over the years have been quite beneficial in this regard. The modality involves the use of micro- and nano-bubble contrast agents and specialized imaging techniques. It permits the visualization of not only the macro- and micro-vasculature but also the measurement of perfusion parameters, which in turn leads to improved diagnostic performance[10]. Unlike CT and MRI, the rapid elimination of the active compounds of CEUS by exhalation allows repeated examinations at short intervals[1]. The fast clearance through the lungs is also particularly helpful for patients with renal insufficiency.

Radiomics

Radiomics is an emerging field within medical imaging that aims to extract mineable quantitative data from routine medical images, such as those obtained from CT, MRI, PET, and in Yang et al’s study[4] CEUS[12,13]. The process typically involves identifying and segmenting a region of interest (ROI) such as a tumor, tumor subregion, or peritumoral zone. The pancreas was ROI for Yang et al’s study[4]. Extracted features typically describe the distribution of signal intensities and spatial relationship of pixels within the ROI. Quantitative CEUS parameters such as maximum intensity, rise time, rise slope 50%, rise slope 10%-90% are then collected and compared with morphologic parameters for pancreatic cancer[3]. All in all, based on the ease of deployment of CEUS (e.g., can be performed at the bedside or other point of care), it is anticipated that there is going to be increased utilization of this imaging modality going forward. It is therefore important to validate its diagnostic capabilities, and this was one of the key goals of Yang et al’s study[4].

Validation of CEUS for pancreatic cancer

Pancreatic neuroendocrine neoplasms: Yang et al[4] uses the terms pancreatic cancer, pancreatic adenocarcinoma and PDAC almost interchangeably throughout the text. This can be problematic since pancreatic cancer includes other entities such as pancreatic neuroendocrine tumors and pancreatic neuroendocrine carcinomas (panNECs), collectively referred to as pancreatic neuroendocrine neoplasms (panNENs)[14-16]. Coincidentally, it is with panNENs that Ki67 (together with mitotic activity) is officially recognized for grading purposes by both the North American Neuroendocrine Tumor Society and European Neuroendocrine Tumor Society as well as the World Health Organization (WHO) and other oncology organizations[17-21]. Indeed, the advent of Ki67 immunostaining was an important milestone for the classification of panNENs because, unlike conventional carcinomas, architectural patterns and cytologic features such as nuclear anisocytosis and atypia (e.g., pleomorphic neuroendocrine tumors) are unreliable for predicting tumor behavior[22]. Ki67 also correlated well with mitotic activity to the extent that both were incorporated into the WHO classification. To this point, the validation of Ki67/mitoses was based on survival outcomes from multiple fairly large studies in different jurisdictions[23-27]. In effect, overall survival (OS) was the gold standard. Viewed from this perspective, Ki67 is a mere parameter, one amongst many, that can be used to assess the biologic aggressiveness of neuroendocrine neoplasms (NENs). Thus, using Ki67 instead of survival makes it a surrogate, not the true gold standard. Moreover, it should also be recognized that Ki67 has its own shortcomings, not least of which are the grading cut-offs and predictive values. This has led a number of studies to question the utility of Ki67 for panNENs[28-30]. It would therefore appear more objective for radiomic studies to use survival as the real gold standard.

Of the panNENs, panNECs are the ones that draw comparisons with PDAC. Molecular profiling shows that genetically and phenotypically, they resemble PDAC in that they carry similar mutations such as TP53, RB1 and KRAS[14,31]. The prognosis for both is very poor with a 5-year OS of < 20%[14]. In contrast, pancreatic neuroendocrine tumours are characterized by MEN1, DAXX/ATRX and mammalian target of rapamycin pathway genes, are slow-growing and have 5- and 15-year OS rates of 85% and 55% respectively[14,25]. Furthermore, it should also be noted that the Ki67 proliferation index for panNECs is defined by a cutoff of > 20%, which is much lower than the 50% threshold that was used in Yang et al’s study[4].

Pancreatic carcinoma subtypes: It is well-accepted that PDAC not otherwise specified (NOS) is the dominant pancreatic cancer. Nonetheless, there are also other variants that may have different disease biology e.g., PDAC arising from intraductal papillary mucinous neoplasm or from mucinous cystic neoplasm, and medullary carcinoma[14]. For example, even though medullary carcinoma is poorly/undifferentiated, it still has a relatively better prognosis compared to PDAC (NOS). Likewise, PDAC ex- intraductal papillary mucinous neoplasm or ex- mucinous cystic neoplasm have better outcomes compared to PDAC (NOS)[14]. There are also mixed carcinomas which can include portions of any of various subtypes. From the perspective of validating a diagnostic modality, it is important to clarify the selection criteria by indicating which of these tumors were included/excluded e.g., colloid carcinoma or acinar carcinoma.

Histologic grading of PDAC: The histologic grading of PDAC is quite different from that of PanNENs. It is based on the extent of glandular differentiation (Figures 1 and 2A) along similar lines to other digestive system adenocarcinomas e.g., in the esophagus, stomach and colorectum[32]. Another grading system uses a combination of degree of glandular differentiation, mucin elaboration, mitoses and nuclear features but overall the predictive value of both these systems is comparable[33]. These systems have been endorsed by official organizations such as the WHO and College of American Pathologists[14,15]. Using this system, most PDACs are well to moderately differentiated, which contrasts with Yang et al’s findings[4] in which almost 50% of the tumors were poorly differentiated. The commonly used guidelines (e.g., WHO and College of American Pathologists) do not specifically mention Ki67 for grading PDAC because it does not necessarily correlate with the histologic grade[14,15,33] (Figure 2B). For this reason, Ki67 is not routinely used as a standard-of-care biomarker for the reporting of PDAC in day-to-day pathology practice. However, those studies that have used it show cut-offs ranging from 10% to 50%[34,35].

Figure 1 Photomicrographs of well-differentiated pancreatic ductal adenocarcinoma. A: Low power view of fine needle biopsy that was obtained using endoscopic ultrasound. The glass slide showed > 2000 pancreatic ductal adenocarcinoma tumor cells which is very helpful for ancillary studies (original magnification × 20); B: Higher power view of one of the cores from Figure 1A showing a well-differentiated pancreatic ductal adenocarcinoma (the well-differentiated glands can be seen on the left and an islet of Langerhans on the right) (original magnification × 100).

Figure 2 Photomicrographs of poorly-differentiated pancreatic ductal adenocarcinoma. A: Section from a pancreaticoduodenectomy specimen showing poorly differentiated pancreatic ductal adenocarcinoma of the pancreatic head. There are poorly formed infiltrative glands in a desmoplastic stroma (original magnification × 100); B: Ki67 immunohistochemistry slide from same case. The positive tumor cell nuclei are brown whereas the negative are blue. The Ki67 Labelling index in this field is 9.4%. Most studies for pancreatic ductal adenocarcinoma use the average count, but for pancreatic neuroendocrine neoplasms, the guidelines mandate selecting the hotspots.

Ki67 and radiomics: Ki67 is a nuclear protein antigen marker that reflects cellular proliferation. Yang et al[4] posits that this mostly occurs in tumors but rarely in normal cells. However, it should be noted that cellular proliferation is part of normal physiologic processes especially in labile cells that continuously regenerate throughout life, e.g., the epidermis, hematopoietic stem cells, and the crypt cell compartment of the mucosal lining as in the gastrointestinal tract[36,37]. In this regard, normal tonsil is commonly used as a positive control in immunohistochemistry protocols for Ki67 not only for staining the germinal centres but also the overlying epithelium[38-40].

From an oncologic perspective, higher Ki-67 expression is often seen as connoting poorer outcomes, suggesting its potential to predict disease aggressiveness and guide treatment approaches[13,41]. As such, it has been used in many radiomics studies as a prognostic marker. However, tumor proliferation is not always the most important indicator of biologic aggressiveness. With some tumors, there may well be other important considerations. For example, tumor budding in PDAC[42], perineural invasion e.g., head and neck carcinomas[43], extramural vascular invasion in jejunoileal neuroendocrine tumors[44], cuproptosis in kidney renal clear cell carcinoma[45], exhausted CD8+ T cells in breast cancer[46], overexpression of the mitogen-activated protein kinase signaling pathway in gliomas[47], or epithelial mesenchymal transition with sarcomatoid carcinomas[14]. All this is to simply say that cell proliferation kinetics should not necessarily be regarded as the prime driver of biologic aggressiveness.

There are also issues relating to the definition of Ki67 threshold values for low- and high-risk carcinoma groups in such sites as the breast, lung and prostate. For example, the cutoffs established by various laboratories and expert committees range from 10% to 50%[34,41,48-50]. These differing Ki67 cut-off values indicate the difficulty in establishing a specific Ki67 cut-off for routine use as a standard-of-care biomarker[51]. What may even be more important is the method that was used for assessing Ki67. For NENs, hotspots are selected for evaluation[14,21-25]. However, in other organs e.g., the breast, lung and prostate, some studies use the average count whereas others use hotspots[35,41,48-50]. Parenthetically, if average counts were used for the WHO classification for NENs, the cut-off would be even lower than 20%.

CEUS vs image-guided biopsies

One of the Yang et al’s main stated objectives in using CEUS was to establish a definitive diagnosis of PDAC while at the same time avoiding the risk of biopsy procedures[4], e.g., damage to major blood vessels in the peripancreatic region. The commonly used biopsy procedures nowadays are image-guided and they are generally considered safe[52,53]. The most widely accepted are percutaneous ultrasound and endoscopic ultrasound biopsy, offering real-time multiplanar scanning capabilities, minimal invasiveness, and a lack of radiation exposure[52-54]. Whereas there may be some form of risk associated with any biopsy procedure, the incidence of complications such as hemorrhage and pancreatitis as well as the occurrence of false-negative results have been considerably lessened over the years due to technical improvements. Another consideration in these comparisons relates to procurement of tissue, either as core biopsies or aspirates[54]. In this respect, CEUS has a major downside since it provides no tissue for diagnostic purposes. As is well known, there are many mimics of PDAC such as chronic pancreatitis, autoimmune pancreatitis, paraduodenal pancreatitis, lymphomas and metastases[55-57]. For a major operation such as Whipple procedure which has considerable morbidity and mortality, it is very important to establish the diagnosis based on firm grounds (Figure 1). Even in inoperable cases, tissue diagnosis is desirable since systemic therapy has significant side-effects. In this day and age of personalized medicine, biopsy also provides material for molecular studies which greatly helps in the selection of the appropriate treatment. At the same time, it is worth noting that there is ongoing research with liquid biopsies and molecular markers which, in the future, could lessen the reliance on traditional tissue biopsies[58].

CEUS vs detective flow imaging-EUS

Ultrasound contrast agents as used in CEUS are among the safest of all contrast media and numerous studies over the years are a testament to their safety profile[59]. Nonetheless, on very rare occasions, they have resulted in hypersensitivity reactions (estimated at 1 in 15000) or even death[11,59]. These issues may possibly explain the initial tardiness of the Food and Drug Administration in approving CEUS, but it was subsequently ratified based on strong multi-societal expert endorsements[59-61]. Partly in response to concerns about hypersensitivity reactions to CEUS agents, detective flow imaging-EUS was developed as a way for detecting fine vessels and low-velocity blood flow without contrast agents. Studies have shown that it can visualize microvascularity much more clearly than color-Doppler EUS, power Doppler EUS or e-FLOW EUS[11]. Therefore, it may be a useful alternative to CEUS for this purpose[11].

CONCLUSION

CEUS is a helpful adjunct to other imaging modalities such as CT, MRI and PET scan, and should be viewed as augmenting rather than replacing routine EUS-guided fine-needle biopsy[62]. As such, both can be performed concurrently or as CEUS-guided EUS-guided fine-needle biopsy[63-65]. This would not only provide more confidence in the diagnosis but also supply material for molecular testing. Future work should use survival outcomes and clear criteria for validation.