Patients and samples
Twenty-seven patients with a total of 56 specimens were enrolled from Sir Run Run Shaw Hospital, College of Medicine, Zhejiang University in this study. Two patients were excluded due to the high content of necrotic tissue in the samples, and one patient was excluded because of insufficient tumor content. Twenty-four MPMTs from 51 specimens met all quality control criteria and were analyzed by the 90-gene expression assay. Based on the invasion site of the tumor, 51 specimens were sorted into five groups: the digestive system (gastroesophagus, colorectum and liver), urinary system (kidney, urinary and prostate), breast, lung, and head and neck (tongue and thyroid). Table 1 presents the demographics of the 24 MPMT patients. Among these patients, 20 harbored synchronous MPMTs, and four harbored metachronous MPMTs. Seventeen patients were male and seven patients were female. The median age at diagnosis was 62.5 (range 45-77) for the first cancer, 63 (range 45-77) for the second cancer and 68 (range 53-70) for the third cancer. The most common invasion sites of the first, second and third tumors were all in the digestive system. Of the 51 specimens, tumors were most frequently located in the colorectum (29.4%, 15 of 51), gastroesophagus (17.6%, 9 of 51), breast (9.8%, 5 of 51) and lung (11.8%, 6 of 51). The distribution of tumor locations is shown in Figure 1. The most common stages of the first, second and third cancers were II, I/II and I/II, respectively. Twenty-three out of 24 patients (95.8%) underwent surgery, and 45.8% (11/24) underwent chemotherapy.
Performance of the 90-gene expression assay in MPMTs
The overall workflow for the 90-gene expression assay is shown in Figure 2. The concentrations of total RNA from 51 samples ranged from 7.3 ng/μL to 214.3 ng/μL, with a median of 66.6 ng/μL. The median A260/A280 ratio (purity of RNA) was 1.98 (range 1.74-2.04).
With the 90-gene expression assay, 16 specimens were classified as colorectal tumors, 9 as gastroesophageal tumors, 6 as breast tumors, 5 as lung tumors, 4 as urinary tumors, 4 as kidney tumors, 2 as liver tumors, 2 as prostate tumors, 2 as thyroid tumors and 1 as head and neck tumors. For the 51 specimens, the 90-gene expression assay showed an overall agreement of 94.1% (48 of 51, 95% confidence interval (CI): 0.83-0.98). Table 2 shows the performance of the 90-gene expression assay in subgroups compared with the reference diagnosis. Twenty-seven digestive system tumors comprised colorectal (n=15), gastroesophageal (n=9) and liver (n=3) tumors. The overall agreement for digestive system tumors was 92.6% (25 of 27, 95% CI: 0.74-0.99), with subgroup accuracy ranging from 66.7% (liver) to 100% (colorectum). In an analysis of the urinary system tumors, the concordance rates were 100% for classifying tumors from the kidney (n=4), urinary (n=4) and prostate (n=2). In addition, the 90-gene expression assay correctly classified 100% of the breast cancer cases (n=5), 100% of the head and neck cancer cases (n=3) and 83.3% of the lung cancer cases (n=6).
To investigate the similarities between clinical samples, hierarchical clustering of the 51 tumors by 90-gene expression profiles was performed. As shown in Figure 3, the hierarchical clustering of 90-gene expression profiling in 51 specimens revealed distinct patterns between the five system types. Samples were grouped together depending on tumor or system types rather than individual MPMT patients.
Three specimens had discordant predictions compared with the reference diagnosis. Detailed clinical information is provided in Table 3. The histological types of the three misclassified samples were pulmonary mucoepidermoid carcinoma, poorly differentiated gastric antrum adenocarcinoma and poorly differentiated gallbladder adenocarcinoma. The 90-gene expression assay misclassified pulmonary mucoepidermoid carcinoma as breast cancer, gastric antrum adenocarcinoma as colorectal cancer, and gallbladder adenocarcinoma as gastroesophageal cancer. For 48 cases that were concordant with the reference diagnosis, the median similarity score was 88 (range 42.8-99.2), while the similarity scores of three misclassified samples were relatively low (38, 49.9 and 56.8). A distribution plot of the highest similarity scores for 51 specimens related to tumor types is displayed in Additional Figure 2.
Specific case
A 77-year-old man noticed chest tightness, shortness of breath, nausea/vomiting and fever. He underwent endoscopic biopsies (29-35 cm from the incisors) and was diagnosed with squamous cell carcinoma (SCC). In the meantime, a chest computed tomography (CT) scan found a lesion in the left upper lung, and the lung biopsy was diagnosed as SCC. Through comprehensive clinical and pathology examination, the clinician confirmed that the patient had multiple synchronous primary tumors according to the Warren and Gates criteria [13]. FFPE tumor tissues taken from the esophagus and lung were analyzed by the 90-gene expression assay, and the predictions showed that the two specimens were gastroesophageal cancer (similarity score = 75.9) and lung cancer (similarity score = 91.7) (Figure 4).
SCC comprises a wide range of tumors originating from diverse anatomical locations that share a common histomorphology and expression of squamous cell differentiation markers, making it difficult to distinguish whether the subsequent SCC is a primary tumor or metastatic lesion. This patient had two lesions on the esophagus and lung simultaneously. Histopathologically, esophageal SCC often metastasizes to the lung. Pathological diagnosis could only confirm the two lesions in the esophagus and lung as SCCs. It was difficult to determine whether the two lesions were synchronous SCCs or represented SCC metastasis based on immunohistochemistry (IHC) and morphology assessment. Clinical outcome of synchronous esophageal and lung SCCs is better than that of metastatic cancer, and the diagnosis of the tumor will directly affect the choice of treatment options. If the new lesion is a second primary tumor, surgical resection supplemented by chemotherapy or radiotherapy has been the preferred therapeutic regimen instead of palliative treatment. Thus, in patients highly suspected of having metastatic cancer, the use of a 90-gene expression assay may be able to identify the tissue of origin more quickly when imaging and IHC examinations are ineffective and could help guide the choice of the precise treatment regimen.