Multi-Site Tumour Sampling Improves the Detection of Intra-Tumour Heterogeneity in Oral and Oropharyngeal Squamous Cell Carcinoma

Background: Oral squamous cell carcinoma (OSCC) and oropharyngeal squamous cell carcinoma (OPSCC) are very common in head and neck malignancy. Intratumour heterogeneity (ITH) may hamper their responses to treatment. Hence, novel tumour sampling methods that reflect ITH are required. In this study, we investigated the clinical significance of multi-site tumour sampling (MSTS) to detect ITH in OSCC and OPSCC. Methods: One hundred eighty-two paired specimens were sampled by routine sampling (RS) or MSTS, respectively. Histologically, tumour grade, peri-tumoural vascular and lymphatic growth, perineural permeation, tumour necrosis, and muscle invasion were assessed. Immunohistochemically, the positive and average detection rates of P53(mutant), ki67 and CyclinD1 were detected. The exon 9 and exon 20 mutations of PIK3CA gene and the methylation status of the CDKN2A promoter were analysed. Results: Microscopically, the detection rate of perineural permeation, the detection density of peri-tumoural vascular and lymphatic growth, necrosis and muscle invasion in MSTS were significantly more frequent than those in RP (P < 0.05, P < 0.05, P < 0.01, P < 0.01). MSTS resulted in a higher detection rate of P53 (mutant), ki67, and CyclinD1 expression than did RS, but the difference was not significant. MSTS's detection rates in PIK3CA gene mutation and gene methylation sequencing in CDKN2A gene promoter region were both higher than RP (P < 0.05, P < 0.01). To be emphasised, the hotspot mutation H1047Rwas detected in one MSTS specimen (case 24M5) but in no RS specimens. Conclusions: This study verified that MSTS's advantage in the reflection of morphological and molecular characteristics of OSCC and OPSCC. MSTS was more representative than RP. Therefore, MSTS can compensate the RP limitations in ITH detection especially in large tumours.


INTRODUCTION
Squamous cell carcinoma (SCC) is a common histological type in oral cavity and oropharynx (1). The mortality rate of these cancers remains poor, owing to its high potential for invasion, maxillofacial destruction, cervical lymph node metastasis, and blood-borne dissemination to distant sites (2). Current treatments include surgery, radiotherapy, chemotherapy, photodynamic therapy, EGFR inhibitors (1), and immunotherapy (3). Although targeted therapies have exerted potent and long-lasting anticancer effects, intratumour heterogeneity (ITH) could lead to treatment failure and therapeutic resistance (4)(5)(6). Mounting evidence have suggested that ITH is extremely complex, hampering the wide success of targeted therapies. Spatial heterogeneity is a form of ITH (7,8). It is common that well-differentiated, poorly differentiated, and moderately differentiated areas can co-exist within an carcinoma. These intra-tumoural differences in differentiation usually lead to varying patterns in lesions, which may cause differential expression of therapeutic targets and influence the treatment response (4).
Genetic heterogeneity always hides behind microscopic homogeneity (4). The PI3K pathway is a well-characterised oncogenic axis (9,10) that has been associated with drug resistance and poor prognosis (11). In head and neck region, PIK3CA is the third most commonly mutated gene (12), and the PI3K signalling is believed to be a key driver of resistance to EGFR inhibitors (13). Tumours with identical histological features may harbour different PI3K mutations. And the heterogeneity in PIK3CA mutations may play a critical role in the response to targeted therapies.
In addition, epigenetic heterogeneity can be often seen in the genetically identical lesions (4). Notably, promoter methylation can cause gene silencing in the absence of DNA mutations (14), leading to differential expression patterns in tumours with the same DNA sequence. For instance, the promoter of CDKN2A is frequently methylated in HNSCC and other oral cancers (15). Differences in the methylation patterns of CDKN2A promoter may affect the treatment response. An accurate pathological diagnosis of OSCC and OPCC is crucial for clinical decision making and treatment response.
Specimen sampling is the cornerstone of accurate diagnosis and clinical decision making. An limitation of routine sampling (RS) practise is that macroscopically homogeneous tumours may differ in their molecular characteristics due to ITH (16). During RS, only a small portion of the tumour is sampled. Thus, the sample may not represent the entire tumour, especially for large tumours (diameter > 3 cm).
Lopez et al. (16) have developed computer mathematical models to improve tumour sampling methods. They proposed a multi-site tumour sampling (MSTS) method based on the computer "divide and conquer algorithm" and found that MSTS can detect ITH in clear cell renal cell carcinoma (CCRCC) efficiently and economically (17)(18)(19)(20). However, the ability of MSTS to detect ITH in OSCC and OPSCC remains unclear (21). In this study, we evaluated the practical usage of MSTS to detect ITH in terms of histology, immunophenotype, mutations, and promoter methylation patterns in OSCC and OPSCC.

Patients and Specimen Collection
OSCC and OPSCC specimens obtained by radical surgery were collected from the Department of Oral Pathology of the Peking University Hospital of Stomatology (Beijing, China). These tumour masses were from 26 cases, and were made pairs, respectively, based on the different sampling method (described following). The inclusion criteria were as follows: (1) SCC (including the subsite of tongue, gingival, buccal, the floor of the mouth, hard palate, base of tongue and oropharynx) was diagnosed by biopsy, (2) maximum tumour diameter of ≥3 cm; (3) provision of informed consent for potential use of the surgical specimens for scientific research. The study was approved by the Ethics Committee of the Peking University Hospital of Stomatology (No. 2017-09-34-21).

Tumour Sampling
To assess the invasion depth, we sampled surgical specimens by excising through the lesion's largest invasion surface with adjacent normal tissues. Specimens were sampled by both RS and MSTS. For RS, after cutting the specimen into parallel cross-sections, we acquired one tissue block as 1.5 cm (length) × 1 cm (width) × 0.3 cm (depth) for each centimetre of mass diameter. For MSTS, we sampled six small scattered tissue blocks [0.5 cm (length) × 0.5 cm (width) × 0.3 cm (depth) each] for each centimetre of mass diameter, and these blocks were used to construct a tissue cassette (Figures 1, 2).

Hematoxylin and Eosin Staining
Specimens obtained by RS and MSTS were embedded in paraffin and processed routinely. Tissue sections were stained with hematoxylin and eosin and reviewed by an experienced pathologist. Tumour grade, inflammation intensity around the tumour, peri-tumoural vascular and lymphatic growth, perineural permeation, tumour necrosis, and muscle invasion were assessed. The detection density of above parameters was calculated as the total detection number/total detection area of tissues. To simplify the calculation, we considered sampling blocks as mathematical shapes (triangle, square, rectangle, trapezoid) and calculated the area of the sampling block using the corresponding formula.

DNA Extraction
DNA extraction was performed using the QIAamp DNA FFPE Tissue Kit (QIAGEN, Dusseldorf, Germany) following the manufacturer's instructions. Particularly, DNA was extracted from six small samples in each MSTS tissue cassette, respectively. DNA quality was analysed by 0.8% agarose gel electrophoresis, and the DNA concentration was assessed using the Nanodrop 2000C spectrophotometer (Thermo Fisher, Waltham, USA).

Mutation Analysis
Exons 9 and 20 of the PIK3CA gene were amplified from genomic DNA by nested polymerase chain reaction (PCR) using a thermocycler (ABI, Carlsbad, USA). The PCR solution contained 100 ng genomic DNA, 10 pmol each primer, 2.5 µL 10 × Taq buffer (with MgCl 2 ), and 1 IU Taq Plus DNA Polymerase (Sangon Biotech, Shanghai, China) in a total volume of 25 µL. Samples were preheated at 95 • C for 5 min, followed by 38 cycles of 94 • C for 30 s, 58 • C for 30 s, and 72 • C for 60 s; a final elongation step was performed at 72 • C for 10 min. The PCR products were sequenced on an autosequencer (ABI, Carlsbad, USA) using the BigDye Terminator Kit (Thermo Fisher, Waltham, USA). Mutations were confirmed by reverse sequencing using the antisense PCR primer and by sequencing a second PCR product from the same sample. The sequences of the primers used for nested PCR and sequencing are shown in Table 1.

Methylation Analysis
The methylation status of the CDKN2A promoter was analysed by bisulfite sequencing PCR. The bisulfite reaction was performed using the EZ DNA Methylation-Gold Kit (ZYMO RESEARCH, Beijing, China) according to the manufacturer's instructions. The bisulfite-modified DNA was amplified by PCR using specific primers. The PCR solution contained 20-50 ng

PIK3CA
Forward primer sequence

Statistical Analysis
IBM SPSS Statistics 20.0 software was used for all statistical analyses. Differences in histological parameters between the sampling methods were analysed by Student's t-test. Differences in the detection rates of different histological parameters, IHC staining results, gene mutations, and methylation status were analysed by Fisher's exact test. Multiple comparisons were performed to determine differences in the average positivity rate of IHC staining among multi-site samples. P < 0.05 were considered statistically significant.

Patient Characteristics
One hundred eighty-two OSCC and OPSCC specimens with a maximum tumour diameter of ≥3 cm were collected from 26 patients (20 men and 6 women). The median age of the patients was 61 (range, 35-79) years. The most common site of involvement was the mobile tongue and gingiva (23.1%), followed by the floor of the mouth (19.2%), oropharynx (15.4%), and base of the tongue (7.7%). The average maximum tumour diameter was 4.2cm (range, 3-7 cm).

IHC Characteristics
RS resulted in a higher p53 detection rate than did MSTS, although the difference between the sampling methods was not significant   difference in the average positivity rate among the multi-site sampling blocks (P < 0.05).

Mutation Detection
All samples were sequenced to detect mutations in PIK3CA exons 9 and 20, which are mutation hotspots. A total of 18 mutation sites were detected in eight MSTS specimens; however, 2 mutation sites were detected in only one RS specimen (case 13), which were also detected in the respective MSTS specimen (case 13M1). Hence, the mutation detection rate was significantly higher in the MSTS than RS specimens [40% (8/20) vs. 5% (1/20); P < 0.05]. Mutations in both exons were detected in the MSTS specimens, whereas mutations in only exon 20 were detected in the RS specimens. Two multi-site sampling blocks from the same patient were found to have the same mutation (3264C > T; cases: 17M2 and 17M4). The hotspot mutation H1047R, which has been shown to enhance PI3K lipid kinase activity, was detected in one MSTS specimen (case 24M5) but in no RS specimens. We also detected two mutations that did not affect the amino acid sequence (3347T > C and 3419G > A; Table 3, Figure 5).

Methylation Detection
Thirteen CpG sites in the promoter region of the CDKN2A gene were analysed by next-generation sequencing. DNA methylation at these sites was detected in all 20 MSTS specimens tested. Nevertheless, only 11 RS specimens were found to have this methylation pattern; these methylation sites were also detected in the MSTS specimens. Therefore, the detection rate of DNA methylation was significantly higher in the MSTS than RS specimens [100% (20/20) vs. 55% (11/20); P < 0.01; Figure 6]. Lopez et al. developed an MSTS sampling method and demonstrated that it was an efficient and economical method to detect ITH in CCRCC (16). They also compared the Fuhrman grade, cell type (clear vs. granular eosinophilic), and tissue necrosis between samples obtained by MSTS and RS from 38 CCRCC patients. Importantly, MSTS yielded significantly higher rates of detecting high-grade and granular eosinophilic cells compared with RS (18). They also analysed PD-L1 (SP142) expression in 22 paired MSTS-obtained and RS-obtained CCRCC specimens and found a higher PD-L1 detection rate with MSTS than RS. Furthermore, MSTS revealed variations in PD-L1 expression within the same tumour (20) Consistent with these findings, the rates of detecting histological and IHC variations in OSCC and OPSCC tumours were higher in MSTS than RS specimens. In this study, we assessed the ability of MSTS to detect ITH in various prognostic indicators, including peri-tumoural vascular and lymphatic growth, perineural permeation, tissue necrosis and muscle invasion. We found a superior performance of MSTS in detecting ITH in all these indicators compared with conventional sampling methods. These results suggested that MSTS improved the accuracy of pathological analysis of tumour samples, and could provide more information of tumour prognosis.

DISCUSSION
ITH in peri-tumoural vascular and lymphatic growth were more evident in the MSTS than RS specimens. Batsakis (24) first described tumour cell infiltration in peripheral nerves. Subsequently, Liebig et al. (25) proposed that tumour cells often infiltrate the nerve sheath. Perineural permeation is predictive of a poor prognosis and has been associated with lymph node involvement, metastasis, local recurrence, and tumour aggression (26)(27)(28)(29). Here, we found a significantly higher rate of perineural permeation detection by MSTS than RS, rendering MSTS particularly useful for the detection of multifocal perineural permeation. Hasmat et al. (30) showed that multifocal perineural invasion was associated with a poor prognosis and higher mortality rate. Park et al. (29) confirmed that multifocal perineural invasion was associated with worse disease-free survival and disease-specific survival. Therefore, by improving the detection rate of multifocal perineural permeation, MSTS may improve patient risk stratification and treatment planning.
Rapid tumour proliferation and insufficient blood supply may enhance tumour necrosis, which has been identified as a predictor of poor prognosis in various solid malignancies, including breast cancer, rectal cancer, lung cancer, renal cell carcinoma, and advanced OSCC (31,32). We found a higher rate of tumour necrosis detection with MSTS than RS.
Chandler et al. (33) proposed that muscle invasion can be used as a surrogate of invasion depth to assess lymph node involvement and the risk of local recurrence. Importantly, our results suggest that MSTS is superior to RS in terms of detecting muscle invasion in OSCC and OPSCC.
was associated with poor prognosis in patients with advanced OSCC. Although we found no significant differences in the rate of p53 (mutant) positivity between the sampling methods, we detected a significant difference in the average positivity among multi-site blocks.
Ki67 is a widely used prognostic marker in numerous cancers, including breast, lung, prostate, and cervical cancers, as well as central nervous system malignancies (39). Evidence suggests that the Ki67 expression level predicts patient prognosis. Jing et al. (40) found that a high Ki67 expression level in the tumour core was associated with poor patient survival. Although we found no significant difference in the rate of Ki67 expression detection between MSTS and RS specimens, we observed differences in the average Ki67 positivity among multi-site blocks, indicating significant ITH in Ki67 expression.
Cyclin D1 is a crucial cell cycle regulator that promotes DNA synthesis and cell division by activating Cdk4 and Cdk6. Khan et al. (41) reported that the expression level of cyclin D1 in OSCC was associated with tumour size, lymph node involvement, clinical stage, as well as patient survival. High expression levels of cyclin D1 and p53 were associated with poor prognosis in patients with advanced OSCC, and co-expression of cyclin D1 and p53 was an independent prognostic factor for OSCC (38). Although we found no significant differences in the proportion of cyclin D1 positivity between the sampling methods, we detected a significant difference in the average positivity among multisite blocks.
We also analysed the ability of MSTS to improve the detection rate of PIK3CA mutations and CDKN2A methylation. The PIK3CA gene is located on chromosome 3q26.3, and its mutation rate in OSCC ranges between 6 and 20% (42). Most PIK3CA mutations are found in exons 9 and 20, which encode the helical region and kinase region domain of PI3K, respectively (43). Here, we showed a significantly higher detection rate of PIK3CA mutations by MSTS than RS. In contrast to RS, which detected mutations only in exon 20, mutations in both exons were detected by MSTS. Importantly, the hotspot mutation H1047R was detected in the MSTS specimens. These results suggest that spatial variation exists in PIK3CA mutations, and that MSTS is superior to RS in detecting PIK3CA mutations. Methylation of tumour suppressor genes is frequent in cancer (44). Notably, CDKN2A promoter methylation is well-documented in head and neck malignancies (15). DNA promoter methylation would cause the silence of CDKN2A gene. CDKN2A regulates of cell division and apoptosis to take part in tumourigenesis, and decelerates cell cycle progression at G1/S phase to maintain the cellular homeostasis (45). CDKN2A silencing due to DNA methylation promotes cell proliferation and tumour development. CDKN2A gene methylation has been reported in 23-67% of primary OSCC tumours (15). In this study, we found a markedly higher detection rate of CDKN2A promoter methylation by MSTS than RS.

CONCLUSION
We applied MSTS to prepare OPSCC specimens for the first time and found that MSTS was superior to RS in detecting ITH in OSCC and OPSCC, including variations in histology, gene expression, clinicopathological characteristics, and mutational and epigenetic landscapes.

DATA AVAILABILITY STATEMENT
The datasets used and analyzed during the current study may be made available by the corresponding author upon reasonable request.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by the Ethics of Committee of the Peking University Hospital of Stomatology. The patients/participants provided their written informed consent to participate in this study.

AUTHOR CONTRIBUTIONS
B-bL initiated and designed the study and revised the manuscript. WJ did the experiments, collected and analysed the data, and drafted the manuscript. JB and JY did the experiments. YC collected the data. All authors approved the final version.