Clinical value of (18)F-FDG PET-CT in detecting primary tumor for patients with carcinoma of unknown primary

Objective: To investigate the clinical value in detecting occult primary tumors with (18)F-FDG PET-CT whole body imaging. Methods: 120 patients with unknown primary origin were referred for (18)F-FDG PET-CT whole body imaging. All patients were performed (18)F-FDG PET-CT whole body scan. PET-CT images were interpreted by visual inspection and semi-quantitative analysis (standardized uptake value, SUV). Histopathological and formal clinical follow-up findings were used to assess the value of FDG PET-CT. Results: FDG PET-CT was able to detect the primary tumor in 54/120 patients (42.5%). The primary tumors were confirmed by histopathologic and formal clinical follow-up findings, and located in the head and neck (n=17), the lung (n=19), the breast (n=2), the esophagus (n=1), the stomach (n=2), the bile ducts (n=1), the pancreas (n=3), the co1on (n=3), the ovary (n=2), the prostate (n=l), others (n=3). FDG PET results were proved false positive in 9 patients (7.5%), which were located in the head and neck (n=3), the lung (n=1), the gastric (n=1), the colon (n=2), the ovary (n=1), the prostate (n=l). During the clinical follow-up of median 32months (range, 2-45months), primary tumor was found in only 5 patients of 60 cases unidentified by PET-CT (breast cancer, gastric cancer, co1on cancer, prostate cancer and urinary tumors, respectively). The sensitivity, specificity, and accuracy of (18)F-FDG PET-CT in the detection of the primary tumor site were 91.5%, 85.2%, and 88.3%, respectively. Conclusion: (18)F-FDG PET-CT whole body imaging is both a noninvasive and a very sensitive tomographic whole-body imaging modality, allowing for the detection of a primary tumor and complete tumor staging in single examination, which can contribute substantially to selecting appropriate therapeutic methods and evaluating prognosis. Perhaps (18)F-FDG PET-CT whole body imaging should be used as a first-line imaging modality for patients with carcinoma of unknown primary rather than using it after other diagnostic procedures have failed to identify a primary tumor.     

Investigating multi-radiomic models for enhancing prediction power of cervical cancer treatment outcomes

Quantitative image features, also known as radiomic features, have shown potential for predicting treatment outcomes in several body sites. We quantitatively analyzed ¹⁸Fluorine–fluorodeoxyglucose (¹⁸F-FDG) Positron Emission Tomography (PET) uptake heterogeneity in the Metabolic Tumor Volume (MTV) of eighty cervical cancer patients to investigate the predictive performance of radiomic features for two treatment outcomes: the development of distant metastases (DM) and loco-regional recurrent disease (LRR). We aimed to fit the highest predictive features in multiple logistic regression models (MLRs). To generate such models, we applied backward feature selection method as part of Leave-One-Out Cross Validation (LOOCV) within a training set consisting of 70% of the original patient cohort. The trained MLRs were tested on an independent set consisted of 30% of the original cohort. We evaluated the performance of the final models using the Area under the Receiver Operator Characteristic Curve (AUC). Accordingly, six models demonstrated superior predictive performance for both outcomes (four for DM and two for LRR) when compared to both univariate-radiomic feature models and Standard Uptake Value (SUV) measurements. This demonstrated approach suggests that the ability of the pre-radiochemotherapy PET radiomics to stratify patient risk for DM and LRR could potentially guide management decisions such as adjuvant systemic therapy or radiation dose escalation.     

肺部病灶18F-FDG PET-CT显像与CT引导下穿刺活检结果对比分析

目的:探讨不同大小病灶18F-FDG PET-CT的显像特点与同层面CT引导下穿刺活检结果的关系.方法:回顾性研究经过18F-FDG PET-CT及CT引导下穿刺活检两种检查的病例,通过统计病灶的大小、病理、穿刺取材部位的放射性浓聚程度、SUVmax 来分析不同大小病灶18F-FDG PET-CT的显像特点与同层面CT引导下穿刺活检结果的关系.结果:69例患者,共穿刺74个痛灶,71个病灶测得SUVmax,放射性浓聚程度不同的病灶其穿刺敏感性、病灶大小的分布情况有显著性差异(P＜0.05),≤ 3cm组病灶放射性均匀浓聚率高,≥5cm组放射性均匀浓聚率低(P＜0.05),≤ 3cm的均匀浓聚的恶性病灶其SUVmax低于＞3 cm组(P＜0.05),良恶性病灶SUVmax无显著性差异(P＞0.05).结论:不同大小的病灶在18F-FDG PET-CT上放射性浓聚程度不同,反映了病灶生物学特性的空间差异,进而影响穿刺活检的准确率.     

Association of mortalin (HSPA9) with liver cancer metastasis and prediction for early tumor recurrence

Hepatocellular carcinoma (HCC) is well known for poor prognosis and short survival because of high recurrence rate even after curative surgery. Today there is no available biomarker or biochemical test to indicate HCC recurrence, and this study aims to identify protein markers that can discriminate postoperative patients with early recurrence (ER), i.e. disease relapsed within the first year. In this study, 103 hepatitis B-related HCC patients were recruited, and 68 of them were used for ER-related biomarker discovery study. Proteomic expression patterns of matched tumor and adjacent non-tumor tissues from these patients plus 16 normal liver tissues were delineated by the two-dimensional gel electrophoresis differential profiling method. Significant protein spots were evaluated by hierarchical clustering analysis. SSP4612 that yielded the highest receiver operating characteristic (ROC) curve value for the ER subgroup of HCC was subsequently identified by tandem mass spectrometry, and the corresponding expression patterns were further confirmed by quantitative PCR, Western blot, and immunohistochemistry. Correlation analysis with clinicopathological data was also examined. Proteomic profiling analysis revealed overexpression of mortalin (gene HSPA9) in HCC when compared with the non-tumor and normal liver tissues (area under the curve (AUC) = 0.821). Furthermore, elevated mortalin level was also detected in the ER subgroup of HCC versus the recurrence-free state (where no cancer recurs for >1 year) (AUC = 0.833, sensitivity = 90.9%, specificity = 71.4%). Metastatic HCC cell lines also exhibited higher levels of mortalin and HSPA9 mRNA. Clinically, mortalin overexpression in HCC was closely associated with advanced tumor stages and venous infiltration, having implications for increased malignancy and aggressive behavior. Mortalin (HSPA9) is associated with HCC metastasis and thus suggested as a tumor marker for predicting early recurrence, which may have immediate clinical applications for cancer surveillance after curative surgery.     

A novel algorithm for network-based prediction of cancer recurrence

To develop accurate prognostic models is one of the biggest challenges in “omics”-based cancer research. Here, we propose a novel computational method for identifying dysregulated gene subnetworks as biomarkers to predict cancer recurrence. Applying our method to the DNA methylome of endometrial cancer patients, we identified a subnetwork consisting of differentially methylated (DM) genes, and non-differentially methylated genes, termed Epigenetic Connectors (EC), that are topologically important for connecting the DM genes in a protein-protein interaction network. The ECs are statistically significantly enriched in well-known tumorgenesis and metastasis pathways, and include known epigenetic regulators. Importantly, combining the DMs and ECs as features using a novel random walk procedure, we constructed a support vector machine classifier that significantly improved the prediction accuracy of cancer recurrence and outperformed several alternative methods, demonstrating the effectiveness of our network-based approach.     

Combined approach of perioperative <Superscript>18</Superscript>F-FDG PET/CT imaging and intraoperative <Superscript>18</Superscript>F-FDG handheld gamma probe detection for tumor localization and verification of complete tumor resection in breast cancer

Background

¹⁸F-fluorodeoxyglucose (¹⁸F-FDG) positron emission tomography/computed tomography (PET/CT) has become an established method for detecting hypermetabolic sites of known and occult disease and is widely used in oncology surgical planning. Intraoperatively, it is often difficult to localize tumors and verify complete resection of tumors that have been previously detected on diagnostic PET/CT at the time of the original evaluation of the cancer patient. Therefore, we propose an innovative approach for intraoperative tumor localization and verification of complete tumor resection utilizing ¹⁸F-FDG for perioperative PET/CT imaging and intraoperative gamma probe detection.

Methods

Two breast cancer patients were evaluated. ¹⁸F-FDG was administered and PET/CT was acquired immediately prior to surgery. Intraoperatively, tumors were localized and resected with the assistance of a handheld gamma probe. Resected tumors were scanned with specimen PET/CT prior to pathologic processing. Shortly after the surgical procedure, patients were re-imaged with PET/CT utilizing the same preoperatively administered ¹⁸F-FDG dose.

Results

One patient had primary carcinoma of breast and a metastatic axillary lymph node. The second patient had a solitary metastatic liver lesion. In both cases, preoperative PET/CT verified these findings and demonstrated no additional suspicious hypermetabolic lesions. Furthermore, intraoperative gamma probe detection, specimen PET/CT, and postoperative PET/CT verified complete resection of the hypermetabolic lesions.

Conclusion

Immediate preoperative and postoperative PET/CT imaging, utilizing the same ¹⁸F-FDG injection dose, is feasible and image quality is acceptable. Such perioperative PET/CT imaging, along with intraoperative gamma probe detection and specimen PET/CT, can be used to verify complete tumor resection. This innovative approach demonstrates promise for assisting the oncologic surgeon in localizing and verifying resection of ¹⁸F-FDG positive tumors and may ultimately positively impact upon long-term patient outcomes.

     

Longitudinal analysis of pre- and post-treatment measurements with equal baseline assumptions in randomized trials

For continuous variables of randomized controlled trials, recently, longitudinal analysis of pre- and posttreatment measurements as bivariate responses is one of analytical methods to compare two treatment groups. Under random allocation, means and variances of pretreatment measurements are expected to be equal between groups, but covariances and posttreatment variances are not. Under random allocation with unequal covariances and posttreatment variances, we compared asymptotic variances of the treatment effect estimators in three longitudinal models. The data-generating model has equal baseline means and variances, and unequal covariances and posttreatment variances. The model with equal baseline means and unequal variance–covariance matrices has a redundant parameter. In large sample sizes, these two models keep a nominal type I error rate and have high efficiency. The model with equal baseline means and equal variance–covariance matrices wrongly assumes equal covariances and posttreatment variances. Only under equal sample sizes, this model keeps a nominal type I error rate. This model has the same high efficiency with the data-generating model under equal sample sizes. In conclusion, longitudinal analysis with equal baseline means performed well in large sample sizes. We also compared asymptotic properties of longitudinal models with those of the analysis of covariance (ANCOVA) and t-test.     

A case of heterotopic pancreas mimicking an intestinal neuroendocrine tumor in 18F-DOPA PET-CT and a review of the literature

18-Fluoro-L-dihydroxyphenylalanine positron emission tomography/computed tomography (18F-DOPA PET-CT) is a nuclear medicine imaging modality indicated for the diagnosis and staging of neuroendocrine tumors (NETs), particularly for the midgut tumors, due to its excellent sensitivity and specificity. Its performance for the detection of foregut-derived NETs (duodenopancreas and proximal jejunum) and for the detection of hindgut-derived NETs is poor and inferior to PET somatostatin receptor imaging such as DOTATOC PET-CT and even inferior to somatostatin analog scintigraphy (octreoscan). There are few studies in the literature on heterotopic pancreas (HP) which is a rare entity, and which can be in some cases a false positive in 18F-DOPA PET-CT. We report a case of HP showing focal uptake on 18F-DOPA PET-CT mimicking an intestinal NET. This case suggests that HP should be included as a possible false positive on 18F-DOPA PET-CT.     

Investigating micronucleus assay applicability for prediction of normal tissue intrinsic radiosensitivity in gynecological cancer patients

BackgroundPelvic organs morbidity after irradiation of cancer patients remains a major problem although new technologies have been developed and implemented. A relatively simple and suitable method for routine clinical practice is needed for preliminary assessment of normal tissue intrinsic radiosensitivity. The micronucleus test (MNT) determines the frequency of the radiation induced micronuclei (MN) in peripheral blood lymphocytes, which could serve as an indicator of intrinsic cell radiosensitivity.AimTo investigate a possible use of the micronucleus test (MNT) for acute radiation morbidity prediction in gynecological cancer patients.Materials and methodsForty gynecological cancer patients received 50 Gy conventional external pelvic irradiation after radical surgery. A four-field “box” technique was applied with 2D planning. The control group included 10 healthy females.Acute normal tissue reactions were graded according to NCI CTCAE v.3.0. From all reaction scores, the highest score named “summarized clinical radiosensitivity” was selected for a statistical analysis.MNT was performed before and after in vitro irradiation with 1.5 Gy. The mean radiation induced frequency of micronuclei per 1000 binucleated cells (MN/1000) and lymphocytes containing micronuclei per 1000 binucleated cells (cells with MN/1000) were evaluated for both patients and controls.An arbitrary cut off value was created to pick up a radiosensitive individual: the mean value of spontaneous frequency of cells with MN/1000 ± 2SD, found in the control group.ResultsBoth mean spontaneous frequency of cells with MN/1000 and MN/1000 were registered to be significantly higher in cancer patients compared to the control group (t = 2.46, p = 0.02 and t = 2.51, p = 0.02). No statistical difference was registered when comparing radiation induced MN frequencies between those groups.Eighty percent (32) of patients developed grade 2 summarized clinical radiosensitivity, with great variations in MNT parameters. Only three patients with grade 2 “summarized clinical radiosensitivity” had values of cells with MN/1000 above the chosen radiosensitivity threshold.ConclusionThe present study was not able to confirm in vitro MNT applicability for radiosensitivity prediction in pelvic irradiation.     

Radioprotection by caffeine pre- and post-treatment in the bone marrow chromosomes of mice given whole-body gamma-irradiation.

The effect of caffeine given as pre- and post-treatment in mice exposed to whole-body gamma-irradiation (1.5 Gy 60Co gamma-rays) was studied. The pre-treatment was either acute or chronic. The acute dose (5 mg/kg and 15 mg/kg body weight) was in the form of an injection given intraperitoneally, 30 min before irradiation. The chronic administration was in the form of caffeine solution (4.208 x 10(-3) M and 7.72 x 10(-4) M) contained in the drinking water that mice had had ad libitum access to instead of plain drinking water for 5 weeks prior to radiation exposure. The acute pre-treatment with caffeine reduced the radiation-induced frequency of chromosomal aberrations discernibly, whereas the chronic pre-treatment afforded a much more significant degree of radioprotection. The caffeine post-treatment (5 mg/kg and 15 mg body weight) was given in the form of an intraperitoneal injection to the mice immediately following whole-body gamma-irradiation. It is noted that both post-treatment concentrations of caffeine also significantly reduced the frequency of chromosomal aberrations induced by gamma-rays. These data are briefly discussed in terms of possible mechanistic considerations.     

A 23-gene prognostic classifier for prediction of recurrence and survival for Asian breast cancer patients

We report a 23- gene-classifier profiled from Asian women, with the primary purpose of assessing its clinical utility towards improved risk stratification for relapse for breast cancer patients from Asian cohorts within 10 years’ following mastectomy. Four hundred and twenty-two breast cancer patients underwent mastectomy and were used to train the classifier on a logistic regression model. A subset of 197 patients were chosen to be entered into the follow-up studies post mastectomy who were examined to determine the patterns of recurrence and survival analysis based on gene expression of the gene classifier, age at diagnosis, tumor stage and lymph node status, over a 5 and 10 years follow-up period. Metastasis to lymph node (N2-N3) with N0 as the reference (N2 vs. N0 hazard ratio: 2.02 (1.05–8.70), N3 vs. N0 hazard ratio: 4.32 (1.41–13.22) for 5 years) and gene expression of the 23-gene panel (P=0.06, 5 years and 0.02, 10 years, log-rank test) were found to have significant discriminatory effects on the risk of relapse (HR (95%CI):2.50 (0.95–6.50)). Furthermore, survival curves for subgroup analysis with N0-N1 and T1-T2 predicted patients with higher risk scores. The study provides robust evidence of the effectiveness of the 23-gene-classifier and could be used to determine the risk of relapse event (locoregional and distant recurrence) in Asian patients, leading to a meaningful reduction in chemotherapy recommendations.     

Methylcap-seq reveals novel DNA methylation markers for the diagnosis and recurrence prediction of bladder cancer in a Chinese population

Purpose

There is a need to supplement or supplant the conventional diagnostic tools, namely, cystoscopy and B-type ultrasound, for bladder cancer (BC). We aimed to identify novel DNA methylation markers for BC through genome-wide profiling of BC cell lines and subsequent methylation-specific PCR (MSP) screening of clinical urine samples.

Experimental Design

The methyl-DNA binding domain (MBD) capture technique, methylCap/seq, was performed to screen for specific hypermethylated CpG islands in two BC cell lines (5637 and T24). The top one hundred hypermethylated targets were sequentially screened by MSP in urine samples to gradually narrow the target number and optimize the composition of the diagnostic panel. The diagnostic performance of the obtained panel was evaluated in different clinical scenarios.

Results

A total of 1,627 hypermethylated promoter targets in the BC cell lines was identified by Illumina sequencing. The top 104 hypermethylated targets were reduced to eight genes (VAX1, KCNV1, ECEL1, TMEM26, TAL1, PROX1, SLC6A20, and LMX1A) after the urine DNA screening in a small sample size of 8 normal control and 18 BC subjects. Validation in an independent sample of 212 BC patients enabled the optimization of five methylation targets, including VAX1, KCNV1, TAL1, PPOX1, and CFTR, which was obtained in our previous study, for BC diagnosis with a sensitivity and specificity of 88.68% and 87.25%, respectively. In addition, the methylation of VAX1 and LMX1A was found to be associated with BC recurrence.

Conclusions

We identified a promising diagnostic marker panel for early non-invasive detection and subsequent BC surveillance.     

Angiopoietin 1 promotes tumor angiogenesis and tumor vessel plasticity of human cervical cancer in mice

We have previously shown that following psoralen photoactivation (PUVA treatment) human dermal fibroblasts undergo long-term growth arrest as well as morphological and functional changes reminiscent of cellular senescence [ 1 ]. In the absence of molecular data on what constitutes normal senescence, it has been difficult to decide whether these PUVA-induced changes reflect cellular senescence or rather a mimic thereof. We herein report that PUVA-induced growth arrest, the senescent phenotype with long-term induction of senescence-associated beta-galactosidase, as well as increased expression of matrix metalloprotease-1 are fully reversible at days 100 to 130 post PUVA treatment in four independently tested fibroblast strains. The late returning growth capacity in PUVA-treated fibroblasts is not due to immortalization, as shown by continued lack of telomerase activity, accelerated telomere shortening, and a decrease in overall growth rates in fibroblasts in their regrowing phase post PUVA treatment. Lack of anchorage-independent growth additionally suggests that the cells are also not tumorigenically transformed. Collectively, our data suggest that PUVA-induced changes do not fully reflect replicative senescence but rather represent a long-term transient phenocopy of senescence. The model reported here is particularly suited to elucidating mechanisms underlying long-term transient growth arrest, the related functional changes, and the release of cells thereof.     

A tumor-associated macrophages related model for predicting biochemical recurrence and tumor immune environment in prostate cancer

ObjectiveTo identify tumor-associated macrophages (TAMs) related molecular subtypes and develop a TAMs related prognostic model for prostate cancer (PCa).MethodsConsensus clustering analysis was used to identify TAMs related molecular clusters. A TAMs related prognostic model was developed using univariate and multivariate Cox analysis.ResultsThree TAMs related molecular clusters were identified and were confirmed to be associated with prognosis, clinicopathological characteristics, PD-L1 expression levels and tumor microenvironment. A TAMs related prognostic model was constructed. Patients in low-risk group all showed a more appreciable biochemical recurrence-free survival (BCRFS) than patients in high-risk group in train cohort, test cohort, entire TCGA cohort and validation cohort. SLC26A3 attenuated progression of PCa and prevented macrophage polarizing to TAMs phenotype, which was initially verified.ConclusionsWe successfully identified molecular clusters related to TAMs. Additionally, we developed a prognostic model involving TAMs that exhibits excellent predictive performance for biochemical recurrence-free survival in PCa.     

DNA methylation of apoptosis genes in rectal cancer predicts patient survival and tumor recurrence

Deregulation of the apoptotic pathway, one of the hallmarks of tumor growth and -progression, has been shown to have prognostic value for tumor recurrence in rectal cancer. In order to develop clinically relevant biomarkers, we studied the methylation status of promoter regions of key apoptosis genes in rectal cancer patients, using methylation-sensitive restriction enzymes. DNA was extracted from fresh-frozen tumor tissues of 49 stage I-III rectal cancer patients and 10 normal rectal tissues. The results of this pilot study were validated in 88 stage III tumor tissues and 18 normal rectal tissues. We found that methylation of the intrinsic apoptotic pathway genes Apaf1, Bcl2 and p53 correlated with the apoptotic status (M30) of the tumor. Combined survival analyses of these three genes, based on the number of genes showing high methylation (all low, 1 high, 2 high or all high), showed shorter patient survival and recurrence-free periods with an increasing number of methylated markers. Multivariate analyses showed significant differences for overall survival (p = 0.01; HR = 0.28 (0.09–0.83)), cancer-specific survival (p = 0.004; HR = 0.13 (0.03–0.67)) and distant recurrence-free survival (p = 0.001; HR = 0.22(0.05–0.94)). The shortest survival was observed for patients showing low methylation of all markers, which—as was expected—correlated with high apoptosis (M30), but also with high proliferation (Ki-67). The study of epigenetic regulation of apoptosis genes provides more insight in the tumorigenic process in rectal cancer and might be helpful in further refining treatment regimens for individual patients.     

Evaluation of serum exosomal LncRNA‐based biomarker panel for diagnosis and recurrence prediction of bladder cancer

Exosomes are small membrane vesicles released by many cells. These vesicles can mediate cellular communications by transmitting active molecules including long non‐coding RNAs (lncRNAs). In this study, our aim was to identify a panel of lncRNAs in serum exosomes for the diagnosis and recurrence prediction of bladder cancer (BC). The expressions of 11 candidate lncRNAs in exosome were investigated in training set (n = 200) and an independent validation set (n = 320) via quantitative real‐time PCR. A three‐lncRNA panel (PCAT‐1, UBC1 and SNHG16) was finally identified by multivariate logistic regression model to provide high diagnostic accuracy for BC with an area under the receiver‐operating characteristic curve (AUC) of 0.857 and 0.826 in training set and validation set, respectively, which was significantly higher than that of urine cytology. The corresponding AUCs of this panel for patients with Ta, T1 and T2‐T4 were 0.760, 0.827 and 0.878, respectively. In addition, Kaplan‐Meier analysis showed that non‐muscle‐invasive BC (NMIBC) patients with high UBC1 expression had significantly lower recurrence‐free survival (P = 0.01). Multivariate Cox analysis demonstrated that UBC1 was independently associated with tumour recurrence of NMIBC (P = 0.018). Our study suggested that lncRNAs in serum exosomes may serve as considerable diagnostic and prognostic biomarkers of BC.     

The cervical microbiome over 7 years and a comparison of methodologies for its characterization

Background

The rapidly expanding field of microbiome studies offers investigators a large choice of methods for each step in the process of determining the microorganisms in a sample. The human cervicovaginal microbiome affects female reproductive health, susceptibility to and natural history of many sexually transmitted infections, including human papillomavirus (HPV). At present, long-term behavior of the cervical microbiome in early sexual life is poorly understood.

Methods

The V6 and V6–V9 regions of the 16S ribosomal RNA gene were amplified from DNA isolated from exfoliated cervical cells. Specimens from 10 women participating in the Natural History Study of HPV in Guanacaste, Costa Rica were sampled successively over a period of 5–7 years. We sequenced amplicons using 3 different platforms (Sanger, Roche 454, and Illumina HiSeq 2000) and analyzed sequences using pipelines based on 3 different classification algorithms (usearch, RDP Classifier, and pplacer).

Results

Usearch and pplacer provided consistent microbiome classifications for all sequencing methods, whereas RDP Classifier deviated significantly when characterizing Illumina reads. Comparing across sequencing platforms indicated 7%–41% of the reads were reclassified, while comparing across software pipelines reclassified up to 32% of the reads. Variability in classification was shown not to be due to a difference in read lengths. Six cervical microbiome community types were observed and are characterized by a predominance of either G. vaginalis or Lactobacillus spp. Over the 5–7 year period, subjects displayed fluctuation between community types. A PERMANOVA analysis on pairwise Kantorovich-Rubinstein distances between the microbiota of all samples yielded an F-test ratio of 2.86 (p<0.01), indicating a significant difference comparing within and between subjects’ microbiota.

Conclusions

Amplification and sequencing methods affected the characterization of the microbiome more than classification algorithms. Pplacer and usearch performed consistently with all sequencing methods. The analyses identified 6 community types consistent with those previously reported. The long-term behavior of the cervical microbiome indicated that fluctuations were subject dependent.     

Computed tomography-based radiomic model at node level for the prediction of normal-sized lymph node metastasis in cervical cancer

PurposeRadiomic models have been demonstrated to have acceptable discrimination capability for detecting lymph node metastasis (LNM). We aimed to develop a computed tomography–based radiomic model and validate its usefulness in the prediction of normal-sized LNM at node level in cervical cancer.MethodsA total of 273 LNs of 219 patients from 10 centers were evaluated in this study. We randomly divided the LNs from the 2 centers with the largest number of LNs into the training and internal validation cohorts, and the rest as the external validation cohort. Radiomic features were extracted from the arterial and venous phase images. We trained an artificial neural network (ANN) to develop two single-phase models. A radiomic model reflecting the features of two-phase images was also built for directly predicting LNM in cervical cancer. Moreover, four state-of-the-art methods were used for comparison. The performance of all models was assessed using the area under the receiver operating characteristic curve (AUC).ResultsAmong the models we built, the models combining the features of two phases surpassed the single-phase models, and the models generated by ANN had better performance than the others. We found that the radiomic model achieved the highest AUCs of 0.912 and 0.859 in the training and internal validation cohorts, respectively. In the external validation cohort, the AUC of the radiomic model was 0.800.ConclusionWe constructed a radiomic model that exhibited great ability in the prediction of LNM. The application of the model could optimize clinical staging and decision-making.