ABSTRACT
Objectives
We aimed to establish the tumor cell count and total cell count as surrogate criteria on the basis of the tissue area of biopsy specimen by endoscopic ultrasound‐guided tissue acquisition (EUS‐TA) for comprehensive genomic profiling (CGP) in pancreatic cancer.
Methods
Samples from 114 patients with unresectable pancreatic ductal adenocarcinoma were analyzed. Tumor cells were visually counted, and total cells were quantified with imaging software. Receiver operating characteristic curves determined optimal cutoffs for achieving DNA yield ≥ 200 ng using the OncoGuide NCC Oncopanel.
Results
Thresholds of 2300 tumor cells (AUC 0.653, 95% CI 0.552–0.750) and 5923 total cells (AUC 0.671, 95% CI 0.565–0.778) were identified. For ≥ 2300 tumor cells, sensitivity was 48.7% and specificity 91.2%, with a positive predictive value (PPV) of 92.9%. For ≥ 5923 total cells, sensitivity was 52.5% and specificity 82.4%, with a PPV of 87.5%. A clinically practical threshold of 6000 total cells yielded a sensitivity of 48.8%, a specificity of 82.4%, and a PPV of 86.7%.
Conclusion
Tumor cell counts ≥ 2300 and total cell counts ≥ 6000 may serve as surrogate criteria for tissue area to determine EUS‐TA specimen suitability for CGP in pancreatic cancer. Given that tumor cell count is a more biologically relevant indicator, total cell count should be used as a supplementary indicator.
Keywords: comprehensive genomic profiling, endoscopic ultrasound‐guided tissue acquisition, pancreatic cancer, total cell count, tumor cell count
Ikeda and colleagues proposed tumor and total cell counts as surrogate indicators of comprehensive genomic profiling suitability in pancreatic cancer EUS‐guided tissue acquisition specimens. Tumor cell count was the primary biologically relevant indicator, with total cell count supplementary. Thresholds of ≥ 2300 tumor cells and ≥ 6000 total cells predicted sufficient DNA yield.
1. Introduction
The incidence of pancreatic cancer has been increasing worldwide, ranking as the sixth most common disease worldwide in men and seventh in women [1]. Often detected at an advanced stage, pancreatic cancer has a poor prognosis and is difficult to cure. For advanced or unresectable malignancies, comprehensive genomic profiling (CGP) can be used to identify actionable genetic alterations, guiding the selection of targeted anti‐tumor drugs and leading to more treatment options. The utility of CGP is currently limited in pancreatic cancer; however, pancreatic cancer remains the second most commonly profiled cancer in Japan, following colorectal cancer, demonstrating the importance of CGP in the clinical management of this malignancy [2].
Pathological evaluation of tissue samples is a prerequisite for CGP. The required evaluation criteria include the tumor tissue area and tumor cellularity of the tissue specimen; correspondingly, the quantity of DNA required for analysis differs for each CGP test kit [3, 4, 5]. Currently, endoscopic ultrasound‐guided tissue acquisition (EUS‐TA) is the standard approach for collecting tumor tissue for pancreatic cancer [6, 7, 8]. However, compared with percutaneous or direct endoscopic biopsies, only a small amount of tissue can be collected; therefore, obtaining a sufficient sample for CGP is often challenging [9]. In fact, CGP submission rates for pancreatic cancer specimens collected using fine‐needle biopsy (FNB) via EUS‐TA, a useful collection method, range from 47% to 82%, indicating that a proportion of patients are still unable to benefit from CGP [10, 11, 12, 13, 14, 15, 16]. Pathological factors contributing to these suboptimal submission rates include the nature of pancreatic cancer tissues, which are highly fibrotic, tough, and hard, making collection by aspiration difficult, even under negative pressure [17, 18]. Consequently, only cellular components within the cancer tissue are often collected as cell clusters, rather than the fibrous tissue characteristic of pancreatic cancer.
Therefore, three major issues prevent CGP submission of pancreatic cancer tissue specimens obtained using EUS‐TA. First, the amount of tissue collected is minimal. Second, measuring the tissue area is difficult because the specimens mainly consist of cell clusters, and there is no well‐established criterion for defining their area. Third, as the amount of nucleic acid required for CGP is calculated from the tissue area, it is often difficult to determine the amount of tissue needed, particularly when the primary specimens are small, fragmented materials containing cell clusters. Consequently, we focused on tumor cell count and total cell count as more objective evaluation criteria that may substitute for tissue area, with an overall objective of facilitating the submission of more cases to CGP. The number of tumor cells in tumor tissue has been reported to be a useful indicator for profiling driver genes in lung cancer; however, this has not been explored in pancreatic cancer [19, 20]. Therefore, this study examined the utility of tumor cell count and total cell count in samples obtained from pancreatic cancer as new indicators of suitability for CGP, as alternatives to tissue area.
2. Patients and Methods
2.1. Study Design
This retrospective cohort study used pathological slides of EUS‐TA specimens collected from patients with unresectable pancreatic adenocarcinoma at the National Cancer Center Hospital, Tokyo, Japan. This study was conducted in collaboration with the National Cancer Center Hospital and Sysmex Corporation (Kobe, Japan). The consent requirement was waived for this study because of its retrospective nature and because the included patients were no longer alive, and the study results could not affect their treatment plan. This study was approved by the Institutional Ethics Board of the National Cancer Center Hospital (Study No. 2018–149) and adhered to the STROBE guidelines for reporting clinical observational studies.
2.2. Patients
This study included patients with unresectable pancreatic adenocarcinoma who underwent EUS‐TA at the National Cancer Center Hospital between January 2017 and December 2023. The OncoGuide NCC Oncopanel system (NOP) (Sysmex) was used for CGP. NOP requires blood and tissue samples to distinguish between somatic and germline mutations. Therefore, eligible patients included those with blood samples stored for research purposes in the NCC‐BioBank at the National Cancer Center or those with blood samples collected for NOP submission during health insurance‐covered treatment. Patients for whom blood samples could not be obtained were excluded.
2.3. Specimen Preparation
Fresh EUS‐TA‐obtained tumor tissues were immediately fixed in 10% neutral‐buffered formalin for 6–48 h and subsequently embedded in paraffin. This formalin‐fixed paraffin‐embedded (FFPE) tissue was sliced into 5‐μm sections and stained with hematoxylin and eosin. Ten unstained 5‐μm sections from the same FFPE tissue, together with paired blood samples, were submitted for NOP testing. DNA was extracted, and genetic analysis was performed at the inspection laboratory at Sysmex. When 10 slices were expected to yield insufficient DNA (< 200 ng), up to 30 slices were prepared at the discretion of the pathologist, and DNA yield per 10 slices was calculated. For insurance‐covered treatment, the NOP submission criteria were a tumor cellularity per slide of ≥ 20% and a tissue area of ≥ 4 mm2.
2.4. NOP Testing
NOP is a CGP system that detects alterations in 124 genes. It can evaluate tumor mutation burden, which may predict the efficacy of immune checkpoint inhibitors, and distinguish germline from somatic mutations, which helps identify hereditary cancers [3, 4].
For NOP analysis, the recommended DNA quantity and quality are ≥ 200 ng and ΔΔCq < 2.0. If DNA is < 200 ng but ≥ 50 ng with 2.0 ≤ ΔΔCq ≤ 2.4, testing may still be performed at the discretion of laboratory experts. However, a DNA quantity of 50 ng or more does not necessarily guarantee successful testing, even when ΔΔCq < 2.0. If tumor cellularity is < 20% before extraction, mutations may not be accurately detected, and results are considered reference values.
Genomic DNA is extracted from tumor tissue and matched blood samples of the same patient. Tissue DNA is quantified by NanoDrop and Qubit, and degradation is assessed by qPCR (Agilent NGS FFPE QC Kit). Adequate DNA (50–200 ng) is processed; 200 ng of blood DNA is also analyzed. DNA is fragmented by ultrasound, repaired at the ends of DNA, and ligated with adapters and indexes using the OncoGuideTM NOP kit. Libraries are pre‐amplified by PCR, purified with AMPureBeads, and checked for fragment size and concentration by TapeStation. DNA that passed the TapeStation specification was hybridized with a biotinylated RNA bait library. Captured DNA is isolated using streptavidin magnetic beads, re‐amplified, purified, and checked for fragment size and concentration with TapeStation.
After obtaining the base sequence using a next‐generation sequencer (NextSeq 550Dx System, Illumina Inc., San Diego, CA, USA), the OncoGuide NCC Oncopanel system was used to detect mutations by performing a matched‐pair analysis of the base sequence data from tumor and non‐tumor samples.
2.5. Outcome Measures
The primary endpoints were tumor cell count and total cell count that corresponded with the 200 ng of DNA extracted from tumor tissues required for NOP analysis. To evaluate whether CGP was performed appropriately, the secondary endpoints were the detection rates of KRAS and/or TP53 alterations in cases with a tumor cellularity of ≥ 20%. Regarding tumor cellularity, the interpretation of results differs depending on the allele frequency of the genetic mutation; therefore, we evaluated whether the tumor cellularity was ≥ 20% or < 20%. We did not set any conditions regarding the tumor tissue area because it did not affect the interpretation of genetic alterations. To determine the suitability of the cell count indicators, the tumor cell count per slide was measured visually using microscopy. All pathological assessments, including diagnosis, tumor cellularity, and tumor cell count, were made by two certified Japanese pathologists (NH and YT). Using the microscopy (BX51, Olympus, Tokyo, Japan) at a magnification of x200, the number of cancer cells was counted. These observers were blinded to each other and also not provided with any clinical information about the patients. The cell counts for each slide obtained by the two pathologists were then compared, and when the difference between their counts was less than 20% of the maximum value, the mean of the two was used as the final count; if the difference exceeded 20%, the observers discussed the reasons for the difference and performed recounts until the difference became less than 20%.
The total cell count per slide of the specimen including non‐tumor cells was measured using Patholoscope (MITANI Corp., Tokyo, Japan) after the microscopic images were imported as digital photo files using a NanoZoomer‐XR (Hamamatsu Photonics, Hamamatsu, Japan). This application was calibrated using 10 EUS‐TA pancreatic cancer specimens that were excluded from the scope of this study.
2.6. Statistical Analysis
As this study concept has not been previously reported, the exact sample size could not be calculated. Therefore, we collected as many cases as possible.
We explored the correlation between tumor cell count and total cell count per slide and the quantity of DNA per 10 tissue sample slides with a thickness of 5 μm submitted to NOP. First, the DNA quantity per tumor or total cell was calculated for all cases. The Spearman's rank correlation coefficient was used to evaluate the correlation between DNA content and cell numbers. Receiver operating characteristic (ROC) curves were then constructed to determine optimal cutoff values, defined as the points with the maximum Yoden coefficient. All statistical analyses were performed using EZR (Saitama Medical Center, Jichi Medical University, Saitama, Japan), and p‐values of < 0.05 were considered statistically significant.
3. Results
3.1. Sample Characteristics
A total of 117 patients with unresectable pancreatic ductal adenocarcinoma who underwent EUS‐TA during the study period and also provided blood samples were included in the study. Table 1 summarizes the patient and sample characteristics. The patients had a median age of 65 years, 84 (72%) were male, and the median tumor size was 33 mm. The primary lesion was the biopsy target in 112 patients (96%), of which 70 (60%) involved the pancreatic head. Metastatic lesions were targeted in 5 patients (4%). The samples were collected by FNB in 92 patients (79%), and the needle gauges were 19G in 54 patients (46%) and 22G in 63 (54%). The puncture route was transgastric in 78 patients (67%), and the median number of passes was 4.
TABLE 1.
Characteristics of patients and biopsy.
| Characteristics | n = 117 |
|---|---|
| Age, years | |
| Median (range) | 65 (39–81) |
| Sex, n (%) | |
| Male | 84 (72) |
| Female | 33 (28) |
| Tumor size, mm | |
| Median (range) | 33 (9–83) |
| Puncture lesion, n (%) | |
| Pancreas | 112 (96) |
| Head of the pancreas | 42 (36) |
| Body and tail of the pancreas | 70 (60) |
| Metastatic lesion | 5 (4) |
| Liver | 3 (2) |
| Lymph node | 2 (2) |
| Needle type, n (%) | |
| EUS‐FNB | 92 (79) |
| EUS‐FNA | 25 (21) |
| Needle size, n (%) | |
| 19G | 54 (46) |
| 22G | 63 (54) |
| Route of puncture, n (%) | |
| Transgastric | 78 (67) |
| Transduodenal | 39 (33) |
| Number of punctures, times | |
| Median (range) | 4 (1–6) |
Abbreviations: EUS, endoscopic ultrasound‐guided; FNA, fine‐needle aspiration; FNB, fine‐needle biopsy; G, gauge.
Of the 117 patients, 84 (72%) had a tumor cellularity of ≥ 20% before NOP submission. Regardless of the number of unstained tissue sections submitted, the median quantity of DNA was 748 ng. When calculating the quantity for 10 slides, the median was 373.95 ng. The total quantity of DNA was ≥ 200 ng in 105 patients (90%), ≥ 50 ng but < 200 ng in 7 (6%), and < 50 ng in 5 (4%). The value of ΔΔCq was < 2.0 in 105 patients (90%), 2.0–2.4 in 2 (2%), and > 2.4 in 7 (6%); the value of ΔΔCq was unknown in 3 patients (2%). A total of 106 patients (91%) met the criteria for NOP analysis on the basis of the quantity and quality of DNA (Table S1 and Figure 1).
FIGURE 1.
Quantity and quality of DNA in collected samples. A total of 106 patients (91%) met the criteria for NOP analysis on the basis of the quantity and quality of DNA. NOP, NCC OncoPanel.
Figure 2 shows the quantity of DNA per cell in all 117 patients, and the median quantity of DNA per cell was 5.96 pg.; regardless, we examined the tumor cell count and total cell count in 114 patients, excluding 3 in which DNA could not be extracted, with DNA amounts of 0 pg.
FIGURE 2.
Quantity of DNA per cell. A box plot was drawn for all cases and outliers were calculated; 3 cases with 0 ng were excluded, resulting in 114 cases to be examined for correlation between cell counts and DNA.
3.2. Optimal Cell Count Cutoff Values
Figures S1 and S2 show the association of the quantity of DNA per 10 slides with tumor cell and total cell count per slide, respectively. Since neither cell counts were normally distributed, we performed a log transformation of cell counts and DNA yield per 10 slides. The tumor cell count ranged from 5 to 11 600 cells per slide, and the correlation coefficient with the quantity of DNA per 10 slides was 0.415 (95% confidence interval [CI]: 0.159–0.482, p < 0.01). The total cell count per slide ranged from 676 to 27 127 cells, and the correlation coefficient with the quantity of DNA per 10 slides was 0.411 (95% CI: 0.284–0.802, p < 0.01). In the ROC analysis, the cutoff values for tumor cell counts and total cell counts were 2300 tumor cells (area under the curve [AUC] 0.653, 95% CI: 0.552–0.750) and 5923 total cells (AUC 0.671, 95% CI: 0.566–0.778), respectively (Figures 3 and 4). Regarding interobserver reproducibility, when two pathologists independently classified cases according to whether the tumor cell count was ≥ 2300 or < 2300, the Cohen's kappa coefficient was 0.68.
FIGURE 3.
Receiver operating characteristic curve for distinguishing the number of tumor cell counts equivalent to 200 ng of quantity of DNA in 114 samples. The cutoff values for tumor cell counts were 2300 (area under the curve: 0.653, 95% confidence interval: 0.552–0.750). AUC, Area under the curve; ROC, Receiver operating characteristic curve.
FIGURE 4.
Receiver operating characteristic curve for distinguishing the number of total cell counts equivalent to 200 ng of quantity of DNA in 114 samples. The cutoff values for total cell counts were 5923 (area under the curve: 0.671, 95% confidence interval: 0.565–0.778).
AUC, Area under the curve; ROC, Receiver operating characteristic curve.
We subsequently conducted a subgroup analysis to eliminate variations in the amount of DNA per cell. The ideal quantity of DNA per cell is reported to be approximately 6 pg [21], whereas the quantity of DNA per cell exceeds 60 pg., showing a particularly wide range in Figure 2. After adjusting the amount of DNA per cell to a more ideal level, 2300 tumor cells (AUC 0.773, 95% CI: 0.657–0.872) and 5923 total cells (AUC: 0.794, 95% CI: 0.690–0.893) were optimal even in a subgroup of 0–10 pg. per cell (Figures S3 and S4). The cutoff values for tumor cell and total cell counts were consistent at 2300 cells and 5923 cells, respectively, in other subgroups (Table S2). A threshold of 2300 tumor cells yielded a sensitivity of 48.7%, a specificity of 91.2%, and a positive predictive value (PPV) of 92.9%. On the other hand, a cutoff of 5923 total cells demonstrated a sensitivity of 52.5%, a specificity of 82.4%, and a PPV of 87.5%. For the purpose of clinical feasibility, we evaluated a total cell count of 6000, which resulted in a sensitivity of 48.8%, a specificity of 82.4%, and a PPV of 86.7% (Figure 5). DNA quantity and quality, along with tumor and total cell count, were analyzed according to needle size and type (Tables S3 and S4). Furthermore, with 19G FNB needles, the median tumor and total cell counts were 2700 and 5580, respectively, whereas with 22G FNB needles, the corresponding median values were 1300 and 4623.
FIGURE 5.
Performance at proposed cell count cutoffs. A threshold of 2300 tumor cells resulted in a sensitivity of 48.7%, a specificity of 91.2%, and a positive predictive value (PPV) of 92.9%. We verified a total cell count of 6000, which is simpler for clinical use; sensitivity, specificity, and PPV were 48.8%, 82.4%, and 86.7%, respectively. PPV, positive predictive value.
3.3. Genetic Data Derived From NOP
In 106 of the 117 patients (91%), the quantity and quality of DNA were considered sufficient for genetic analysis. For the remaining 11 patients, the quantity of DNA was < 50 ng or the ΔΔCq value was > 2.4, and genetic analysis was impossible (Figure 1). Of the 117 patients, 7 had a ΔΔCq value of > 2.4, with 4 of these being samples that had been stored for > 3 years. For the other 3 patients, the quantity of DNA was > 200 ng, but the tumor cell count was < 2300 cells; in one patient, the tumor cellularity was < 20%. In addition, 5 of the 117 patients had DNA quantities of < 50 ng. Of these, three had a tumor cell count of ≤ 2300 cells, and the tumor cellularity was < 20%.
Of the 106 patients who were profiled, 79 (75%) had a tumor cellularity of ≥ 20% before NOP submission. The major pathogenic mutations identified in these 79 patients are listed in Table 2. Overall, 71 (90%), 60 (76%), 18 (23%), and 12 (15%) patients tested positive for genetic mutations in KRAS, TP53, SMAD4, and CDKN2A, respectively. All patients had at least one mutation in one of these four genes. Genetic analysis was also performed on patients with a tumor cellularity of < 20%. Of these 27 patients, 23 tested positive for KRAS, TP53, SMAD4, or CDKN2A.
TABLE 2.
Major pathogenic alterations identified using NOP.
| Major pathogenic alterations, n | |||
|---|---|---|---|
| KRAS | 71 | GNAS | 3 |
| TP53 | 60 | KDM6A | 2 |
| SMAD4 | 18 | MAP2K4 | 2 |
| CDKN2A | 12 | MYC | 3 |
| ACTN4 | 1 | NOTCH | 1 |
| ARID1A | 1 | NRAS | 1 |
| ARID2 | 1 | NTRK3 | 1 |
| BRAF | 2 | PALB2 | 1 |
| BRCA1/2 | 1/4 | PBRM1 | 1 |
| CCND1 | 1 | PIK3CA | 2 |
| CCNE1 | 1 | PTEN | 1 |
| CDK6 | 1 | RB1 | 1 |
| ENO1 | 1 | SMARCA4 | 1 |
| ERBB2/4 | 1/1 | STK11 | 1 |
| EP300 | 1 | ||
Abbreviation: NOP, NCC OncoPanel.
4. Discussion
To the best of our knowledge, this is the first report showing that tumor cell and total cell counts in a biopsy sample can be used as a surrogate indicator for tissue area, which is the criterion for evaluating pancreatic cancer tissue samples for CGP. The CGP submission rate of pancreatic cancer EUS‐TA specimens is low because of the difficulty in acquiring these samples and the pathological characteristics of pancreatic cancer. Therefore, alternative evaluation criteria are needed more than in other cancers.
Tissue sampling techniques have improved with the development of new needles and puncture methods [10, 11, 12, 13, 14, 15, 16]. In addition, new specimen processing methods for EUS‐TA specimens, such as macroscopic and stereomicroscopy on‐site evaluation, have been implemented to obtain tissue with higher tumor content [22, 23]. However, the evaluation method for histological specimens suitable for CGP, which is necessary before CGP submission, has not sufficiently improved. Studies have shown that tumor cellularity is overestimated in carcinomas, excluding pancreatic cancer, and that this can be improved by specialized training for pathologists [24, 25]. Measuring tumor tissue area histologically is an indirect method of evaluating the total number of tumor cells that are assumed to be present in a certain percentage of the tumor tissue for CGP; consequently, this evaluation method does not account for tumor cell clusters that are detached from the tissue. The validity of these standard methods of evaluation remains uncertain for samples taken from hard fibrotic cancerous tissues such as pancreatic cancer, as these comprise mainly cell clusters rather than tissue components. Therefore, standard evaluation methods likely underestimate the number of tumor cells contained within pancreatic cancer samples obtained by EUS‐TA. We reported that 53.6% of EUS‐TA pancreatic cancer specimens could not be submitted to NOP because of an insufficient tissue area, whereas 31.4% could not be submitted because of an insufficient tumor cellularity. However, the median quantity of DNA in the submitted specimens exceeded the 200 ng required for NOP analysis, making it difficult to compare the tumor cellularity and tissue area with other cancer types; this indicates that the development of evaluation methods is not sufficient for these pathological parameters [14].
When considering the number of cells, tumor cell count has been reported as an evaluation criterion for OncoMine, a companion diagnostic tool for seven important driver genes in lung cancer [19, 20]. For pancreatic cancer, if the tumor cell count is ≥ 2000 cells when using Patholoscope, the EUS‐TA specimen can be considered appropriate for submission to FoundationOneCDx (FOne). However, in the same report, the median number of tumor cells in successful CGP cases was 4987 cells, and the median number of tumor cells in cases considered unsuitable for CGP was 2192 cells. Therefore, cases with ≥ 2000 tumor cells were considered suitable for CGP if tumor cellularity was adequate; however, the appropriateness of 2000 cells as a cutoff value for Fone submission was not validated [26]. The current study revealed that a tumor cell count of 2300 cells and a total cell count of 5923 cells corresponded to the 200 ng of DNA required for NOP analysis. Furthermore, we proposed a cutoff value of 6000 total cells as appropriate for clinical application. This value yielded sensitivity, specificity, and positive predictive value comparable to that of the original thresholds. On the basis of needle‐specific analyses, a 19G FNB needle was suggested to be the most suitable for obtaining a cell quantity close to this proposed cutoff value. Implementing the cutoff values determined in this study is considered cost‐effective because these values make it possible to reliably perform CGP testing, mitigating the loss of costs due to CGP failure. However, the tumor cell count needs to be measured by a pathologist, increasing the costs of labor; likewise, evaluation of the total cell count entails additional financial burden because of the need for implementing software.
We found that all cases with a tumor cellularity of ≥ 20% had genetic alterations in either KRAS or TP53, regardless of the cell count, suggesting that gene analysis using CGP may be conducted accurately with EUS‐TA specimens in pancreatic cancer. Furthermore, if the tumor cellularity was > 20%, NOP analysis was successful in all cases, suggesting that analyzing DNA quantities of < 200 ng is possible. In this study, 96% of the patients had > 50 ng of DNA. If a quantity of DNA of ≥ 50 ng and quality of ΔΔCq of ≤ 2.4 can be guaranteed, CGP can be offered to more cases in the future.
One limitation of this study is its single‐center design. Multiple centers should be involved in further validation, as the single‐center nature of the study may have led to biases in the evaluation and patient population. Second, the ideal quantity of DNA per cell is approximately 6 pg [21]. Consequently, we assumed that the amount of DNA per cell was an ideal value to eliminate variations in subgroup analysis. This variation was likely due to the difference in cellular size (10–15 μm), which may have caused a difference between the number of cells counted using a microscopic slide specimen and the actual number of cells, depending on the number of submitted slides [20]. Although all the sections were obtained from the same tissue block, slight differences between slides could have resulted in variations in the quantity of DNA per cell. Third, this study demonstrated that both tumor cell count and total cell count have limited sensitivity and AUC. That is, even specimens with relatively few cells may still yield more than 200 ng of DNA. This is likely attributable to variability in cell numbers within the specimens, reflecting intratumoral heterogeneity influenced by pancreatic cancer–specific fibrosis and fragmentation characteristic of FNA specimens. Thus, although they may not serve as complete surrogate indicators, if they are applied as supplementary tools, for example, by calculating tumor cell and total cell count in cases judged by pathologists to have insufficient tissue area and thereby allowing the submission of specimens with a small tissue area, they could represent practically meaningful indicators. Furthermore, although total cell count may correspond to a DNA quantity of 200 ng, CGP requires tumor‐derived cells, and tumor cell count may represent a more relevant and necessary requirement than total cell count. It is important to further advance research from the perspective of tumor cell count, which is an important determinant for improving the success of CGP. Fourthly, the appropriate number of cells has not been explored for other CGP tests that use tissues for analysis and have been approved in Japan, such as FOne and GenMineTOP (TOP). FOne requires a tissue area of ≥ 25 mm2, which is larger than the ideal NOP requirement of 16 mm2 and minimum requirement of 4 mm2. TOP has similar tissue area requirements as NOP. For FOne and TOP, the area, thickness, and number of sheets must be multiplied to obtain volumes ≥ 1 mm3 and ≥ 1.3 mm3, respectively. In contrast with NOP, FOne and TOP do not have minimum requirements for quantity of DNA. Therefore, conducting the same analysis as in this study is not possible with FOne and TOP.
In conclusion, this study proposed tumor cell count and total cell count as novel criteria for evaluating EUS‐TA specimens of pancreatic cancer for NOP analysis. We demonstrated that a tumor cell count of ≥ 2300 cells or a total cell count of ≥ 6000 cells provided the required 200 ng of DNA for successful NOP analysis. Importantly, tumor cell count represents the primary and more biologically relevant indicator, whereas total cell count should be regarded as a supplementary factor. These criteria may serve as practical indicators for the challenging assessment of pancreatic cancer specimens and may facilitate the submission of a greater number of EUS‐TA samples for CGP.
Funding
This study was supported in part by Japan Society for the Promotion of Science (JSPS) KAKENHI (Grant Number 24 K02254; 10.13039/501100001691) and Sysmex Corporation. Sysmex Corporation also provided funding for the analysis of cases in which archived specimens were used for NOP analysis.
Conflicts of Interest
Nobuyoshi Hiraoka and Susumu Hijioka received consulting fees from Sysmex Corporation. Susumu Hijioka received honoraria for lectures from AstraZeneca, Bristol‐Myers Squibb, and J‐MIT. The National Cancer Center Hospital received grants or contracts from Syneos Health.
Supporting information
FIGURE S1: Correlation between tumor cell count and quantity of DNA in 10 sheets of sample. Tumor cell counts ranged from 5 to 11,600 cells per slide, and the correlation coefficient with quantity of DNA per 10 slides was 0.415 (95% confidence interval: 0.159–0.482, p < 0.01).
FIGURE S2: Correlation between total cell count and quantity of DNA in 10 sheets of sample. Total cell counts per slide ranged from 676 to 27,127 cells, and the correlation coefficient with a quantity of DNA per 10 slides was 0.411 (95% confidence interval: 0.284–0.802, p < 0.01).
FIGURE S3: Receiver operating characteristic curve for distinguishing the number of tumor cell counts equivalent to 200 ng of quantity of DNA in a subgroup of 0‐10 pg per cell. The cutoff values for tumor cell counts was 2,300 (area under the curve: 0.773, 95% confidence interval: 0.657‐0.872).
FIGURE S4: Receiver operating characteristic curve for distinguishing the number of total cell counts equivalent to 200 ng of quantity of DNA in a subgroup of 0‐10 pg per cell. The cutoff values for tumor cell counts was 5,923 (area under the curve: 0.794, 95% confidence interval: 0.690–0.893).
TABLE S1: Characteristics of tissue sample.
TABLE S2: The number of tumor cell counts and total cell counts equivalent to 200 ng of quantity of DNA using receiver operating characteristic curve in subgroup analysis.
TABLE S3: Results stratified by needle size.
TABLE S4: Results stratified by needle type.
Acknowledgments
We would like to thank the members of the National Cancer Center Hospital, Japan, and Sysmex for their support of this work. We thank Ms. Sachiko Miura, Kiyono Uchida, and Yumiko Sato for their excellent techniques. We are grateful to the National Cancer Center Biobank for the blood samples used in this study.
Contributor Information
Susumu Hijioka, Email: shijioka@ncc.go.jp.
Nobuyoshi Hiraoka, Email: nhiraoka@ncc.go.jp.
Data Availability Statement
The data that support the findings of this study are available from the corresponding author upon reasonable request.
References
- 1. Bray F., Laversanne M., Sung H., Ferlay J., Torre L. A., and Jemal A., “Global Cancer Statistics 2022: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries,” CA: a Cancer Journal for Clinicians 74 (2024): 229–263, 10.3322/caac.21834. [DOI] [PubMed] [Google Scholar]
- 2. Kohno T., Kato M., Kohsaka S., Goto K., Mano H., and Nakagawa H., “C‐CAT: The National Datacenter for Cancer Genomic Medicine in Japan,” Cancer Discovery 12 (2022): 2509–2515, 10.1158/2159-8290.CD-22-0417. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Sunami K., Ichikawa H., Kubo T., Kato K., Momozawa Y., and Yamaguchi K., “Feasibility and Utility of a Panel Testing for 114 Cancer‐Associated Genes in a Clinical Setting: A Hospital‐Based Study,” Cancer Science 110 (2019): 1480–1490, 10.1111/cas.13969. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Nishino T., Yumura M., Sunami K., et al., “Leveraging Non‐Coding Regions to Guarantee the Accuracy of Small‐Sized Panel‐Based Tumor Mutational Burden Estimates,” Cancer Science 115 (2024): 4045–4053, 10.1111/cas.16342. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Foundation Medicine , Inc. FoundationOne® CDx Specimen Instructions. Cambridge, MA: Foundation Medicine, Inc.; 2024. Available at: https://www.foundationmedicine.com/sites/default/files/media/documents/2024‐04/F1CDx_Specimen_Instructions.pdf. Accessed October 20, 2025. [Google Scholar]
- 6. Banafea O., Mghanga F. P., Zhao J., et al., “Endoscopic Ultrasonography With Fine‐Needle Aspiration for Histological Diagnosis of Solid Pancreatic Masses: A Meta‐Analysis of Diagnostic Accuracy Studies,” BMC Gastroenterology 16 (2016): 108, 10.1186/s12876-016-0524-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Puli S. R., Bechtold M. L., Buxbaum J. L., and Eloubeidi M. A., “How Good Is Endoscopic Ultrasound‐Guided Fine‐Needle Aspiration in Diagnosing the Correct Etiology for a Solid Pancreatic Mass?: A Meta‐Analysis and Systematic Review,” Pancreas 42 (2013): 20–26, 10.1097/MPA.0b013e3182546e79. [DOI] [PubMed] [Google Scholar]
- 8. Yang Y., Li L., Qu C., Liang S., Zeng B., and Luo Z., “Endoscopic Ultrasound‐Guided Fine Needle Core Biopsy for the Diagnosis of Pancreatic Malignant Lesions: A Systematic Review and Meta‐Analysis,” Scientific Reports 6 (2016): 22978, 10.1038/srep22978. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Okuno N., Hara K., Mizuno N., et al., “Clinical Utility of Endoscopic Ultrasound‐Guided Tissue Acquisition for Comprehensive Genomic Profiling of Pancreatic Cancer,” Clinical Endoscopy 56 (2023): 221–228, 10.5946/ce.2021.248. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Larson B. K., Tuli R., Jamil L. H., et al., “Utility of Endoscopic Ultrasound‐Guided Biopsy for Next‐Generation Sequencing of Pancreatic Exocrine Malignancies,” Pancreas 47 (2018): 990–995, 10.1097/MPA.0000000000001132. [DOI] [PubMed] [Google Scholar]
- 11. Park J. K., Lee J. H., Noh D. H., et al., “Factors of Endoscopic Ultrasound‐Guided Tissue Acquisition for Successful Next‐Generation Sequencing in Pancreatic Ductal Adenocarcinoma,” Gut and Liver 14 (2020): 387–394, 10.5009/gnl19214. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Kandel P., Nassar A., Gomez V., et al., “Comparison of Endoscopic Ultrasound‐Guided Fine‐Needle Biopsy Versus Fine‐Needle Aspiration for Genomic Profiling and DNA Yield in Pancreatic Cancer: A Randomized Crossover Trial,” Endoscopy 53 (2021): 376–382, 10.1055/a-1235-3440. [DOI] [PubMed] [Google Scholar]
- 13. Hisada Y., Hijioka S., Ikeda G., et al., “Proportion of Unresectable Pancreatic Cancer Specimens Obtained by Endoscopic Ultrasound‐Guided Tissue Acquisition Meeting the OncoGuide NCC Oncopanel System Analysis Suitability Criteria: A Single‐Arm, Phase II Clinical Trial,” Journal of Gastroenterology 57 (2022): 990–998, 10.1007/s00535-022-01923-2. [DOI] [PubMed] [Google Scholar]
- 14. Ikeda G., Hijioka S., Nagashio Y., et al., “Fine‐Needle Biopsy With 19G Needle Is Effective in Combination With Endoscopic Ultrasound‐Guided Tissue Acquisition for Genomic Profiling of Unresectable Pancreatic Cancer,” Digestive Endoscopy 35 (2023): 124–133, 10.1111/den.14296. [DOI] [PubMed] [Google Scholar]
- 15. Ishigaki K., Nakai Y., Endo G., et al., “Feasibility of Comprehensive Genomic Profiling Using Endoscopic Ultrasound‐Guided Tissue Acquisition With a 22‐Gauge Franseen Needle,” DEN Open 4 (2024): e365, 10.1002/deo2.365. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Buchberg J., de Stricker K., Pfeiffer P., et al., “Mutational Profiling of 103 Unresectable Pancreatic Ductal Adenocarcinomas Using EUS‐Guided Fine‐Needle Biopsy,” Endoscopic Ultrasound 13 (2024): 154–164, 10.4103/EUS-D-23-00137. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Cancer Genome Atlas Research Network , “Integrated Genomic Characterization of Pancreatic Ductal Adenocarcinoma,” Cancer Cell 32 (2017): 185, 10.1016/j.ccell.2017.07.007. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Wood L. D. and Hruban R. H., “Pathology and Molecular Genetics of Pancreatic Neoplasms,” Cancer Journal 18 (2012): 492–501, 10.1097/PPO.0b013e31826c53f6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Nemoto D., Yokose T., Katayama K., et al., “Tissue Surface Area and Tumor Cell Count Affect the Success Rate of the Oncomine Dx Target Test in the Analysis of Biopsy Tissue Samples,” Thoracic Cancer 12 (2021): 194–200, 10.1111/1759-7714.13757. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Iwama E., Yamamoto H., Okubo F., et al., “Evaluation of Appropriate Conditions for Oncomine DxTT Testing of FFPE Specimens for Driver Gene Alterations in Non‐Small Cell Lung Cancer,” Thoracic Cancer 14 (2023): 2288–2296, 10.1111/1759-7714.14822. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Chen H., Luthra R., Goswami R. S., et al., “Analysis of Pre‐Analytic Factors Affecting the Success of Clinical Next‐Generation Sequencing of Solid Organ Malignancies,” Cancers (Basel) 7 (2015): 1699–1715, 10.3390/cancers7030851. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Iwashita T., Yasuda I., Mukai T., et al., “Macroscopic on‐Site Quality Evaluation of Biopsy Specimens to Improve the Diagnostic Accuracy During EUS‐Guided FNA Using a 19‐Gauge Needle for Solid Lesions: A Single‐Center Prospective Pilot Study (MOSE Study),” Gastrointestinal Endoscopy 81 (2015): 177–185, 10.1016/j.gie.2014.07.047. [DOI] [PubMed] [Google Scholar]
- 23. Okuwaki K., Watanabe M., Yoshida T., et al., “Efficacy of Endoscopic Ultrasound‐Guided Tissue Acquisition Using Stereo‐Microscopic on‐Site Evaluation for Possible Comprehensive Genome Profile in Patients With Advanced Pancreatic Cancer,” Journal of Gastroenterology and Hepatology 39 (2024): 740–745, 10.1111/jgh.16374. [DOI] [PubMed] [Google Scholar]
- 24. Mikubo M., Seto K., Kitamura A., et al., “Calculating the Tumor Nuclei Content for Comprehensive Cancer Panel Testing,” Journal of Thoracic Oncology 15 (2020): 130–137, 10.1016/j.jtho.2019.10.018. [DOI] [PubMed] [Google Scholar]
- 25. Smits A. J. J., Kummer J. A., de Bruin P. C., et al., “The Estimation of Tumor Cell Percentage for Molecular Testing by Pathologists Is Not Accurate,” Modern Pathology 27 (2014): 168–174, 10.1038/modpathol.2013.145. [DOI] [PubMed] [Google Scholar]
- 26. Washimi K., Hiroshima Y., Sato S., et al., “Evaluation of Pancreatic Cancer Specimens for Comprehensive Genomic Profiling,” Pathology International 74 (2024): 252–261, 10.1111/pin.13498. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
FIGURE S1: Correlation between tumor cell count and quantity of DNA in 10 sheets of sample. Tumor cell counts ranged from 5 to 11,600 cells per slide, and the correlation coefficient with quantity of DNA per 10 slides was 0.415 (95% confidence interval: 0.159–0.482, p < 0.01).
FIGURE S2: Correlation between total cell count and quantity of DNA in 10 sheets of sample. Total cell counts per slide ranged from 676 to 27,127 cells, and the correlation coefficient with a quantity of DNA per 10 slides was 0.411 (95% confidence interval: 0.284–0.802, p < 0.01).
FIGURE S3: Receiver operating characteristic curve for distinguishing the number of tumor cell counts equivalent to 200 ng of quantity of DNA in a subgroup of 0‐10 pg per cell. The cutoff values for tumor cell counts was 2,300 (area under the curve: 0.773, 95% confidence interval: 0.657‐0.872).
FIGURE S4: Receiver operating characteristic curve for distinguishing the number of total cell counts equivalent to 200 ng of quantity of DNA in a subgroup of 0‐10 pg per cell. The cutoff values for tumor cell counts was 5,923 (area under the curve: 0.794, 95% confidence interval: 0.690–0.893).
TABLE S1: Characteristics of tissue sample.
TABLE S2: The number of tumor cell counts and total cell counts equivalent to 200 ng of quantity of DNA using receiver operating characteristic curve in subgroup analysis.
TABLE S3: Results stratified by needle size.
TABLE S4: Results stratified by needle type.
Data Availability Statement
The data that support the findings of this study are available from the corresponding author upon reasonable request.