Tumor Marker Data Research Articles

Machine learning prediction models for different stages of non-small cell lung cancer based on tongue and tumor marker: a pilot study

ObjectiveTo analyze the tongue feature of NSCLC at different stages, as well as the correlation between tongue feature and tumor marker, and investigate the feasibility of establishing prediction models for NSCLC at different stages based on tongue feature and tumor marker.MethodsTongue images were collected from non-advanced NSCLC patients (n = 109) and advanced NSCLC patients (n = 110), analyzed the tongue images to obtain tongue feature, and analyzed the correlation between tongue feature and tumor marker in different stages of NSCLC. On this basis, six classifiers, decision tree, logistic regression, SVM, random forest, naive bayes, and neural network, were used to establish prediction models for different stages of NSCLC based on tongue feature and tumor marker.ResultsThere were statistically significant differences in tongue feature between the non-advanced and advanced NSCLC groups. In the advanced NSCLC group, the number of indexes with statistically significant correlations between tongue feature and tumor marker was significantly higher than in the non-advanced NSCLC group, and the correlations were stronger. Support Vector Machine (SVM), decision tree, and logistic regression among the machine learning methods performed poorly in models with different stages of NSCLC. Neural network, random forest and naive bayes had better classification efficiency for the data set of tongue feature and tumor marker and baseline. The models’ classification accuracies were 0.767 ± 0.081, 0.718 ± 0.062, and 0.688 ± 0.070, respectively, and the AUCs were 0.793 ± 0.086, 0.779 ± 0.075, and 0.771 ± 0.072, respectively.ConclusionsThere were statistically significant differences in tongue feature between different stages of NSCLC, with advanced NSCLC tongue feature being more closely correlated with tumor marker. Due to the limited information, single data sources including baseline, tongue feature, and tumor marker cannot be used to identify the different stages of NSCLC in this pilot study. In addition to the logistic regression method, other machine learning methods, based on tumor marker and baseline data sets, can effectively improve the differential diagnosis efficiency of different stages of NSCLC by adding tongue image data, which requires further verification based on large sample studies in the future.

Comprehensive Profiling of Cancer-Associated Cells in the Blood of Breast Cancer Patients Undergoing Neoadjuvant Chemotherapy to Predict Pathological Complete Response.

Neoadjuvant chemotherapy (NAC) can affect pathological complete response (pCR) in breast cancers; the resection that follows identifies patients with residual disease who are then offered second-line therapies. Circulating tumor cells (CTCs) and cancer-associated macrophage-like cells (CAMLs) in the blood can be used as potential biomarkers for predicting pCR before resection. CTCs are of epithelial origin that undergo epithelial-to-mesenchymal transition to become more motile and invasive, thereby leading to invasive mesenchymal cells that seed in distant organs, causing metastasis. Additionally, CAMLs in the blood of cancer patients are reported to either engulf or aid the transport of cancer cells to distant organs. To study these rare cancer-associated cells, we conducted a preliminary study where we collected blood from patients treated with NAC after obtaining their written and informed consent. Blood was collected before, during, and after NAC, and Labyrinth microfluidic technology was used to isolate CTCs and CAMLs. Demographic, tumor marker, and treatment response data were collected. Non-parametric tests were used to compare pCR and non-pCR groups. Univariate and multivariate models were used where CTCs and CAMLs were analyzed for predicting pCR. Sixty-three samples from 21 patients were analyzed. The median(IQR) pre-NAC total and mesenchymal CTC count/5 mL was lower in the pCR vs. non-pCR group [1(3.5) vs. 5(5.75); p = 0.096], [0 vs. 2.5(7.5); p = 0.084], respectively. The median(IQR) post-NAC CAML count/5 mL was higher in the pCR vs. non-pCR group [15(6) vs. 6(4.5); p = 0.004]. The pCR group was more likely to have >10 CAMLs post-NAC vs. non-pCR group [7(100%) vs. 3(21.4%); p = 0.001]. In a multivariate logistic regression model predicting pCR, CAML count was positively associated with the log-odds of pCR [OR = 1.49(1.01, 2.18); p = 0.041], while CTCs showed a negative trend [Odds Ratio (OR) = 0.44(0.18, 1.06); p = 0.068]. In conclusion, increased CAMLs in circulation after treatment combined with lowered CTCs was associated with pCR.

The impact of HIV infection on clinical presentation and mortality among persons with hepatocellular carcinoma in Kampala, Uganda

BackgroundHIV infection is associated with more rapid progression of some comorbidities. This study assessed the impact of HIV-infection on the presentation and outcome of HCC.MethodsHCC patients attending the Mulago National Referral Hospital in Uganda were enrolled into a natural history study of HCC between March 2015 and February 2019. Standardized methods were used to collect clinical, ultrasound and laboratory data at enrolment. HCC cases were confirmed and enrolled based on a combination of clinical, ultrasound, tumor marker and pathology data. Follow-up contact was made at one, three, six, and twelve months post-enrolment to determine vital status. Symptoms and signs at diagnosis and subsequent survival were compared by HIV status. Kaplan Meier curves were used to assess HCC survival.ResultsOf 441 persons with HCC, 383 (87.0%) died within 12 months following HCC diagnosis. The median (IQR) survival was 42 (20, 106) days. HIV infection was present in 79 (18%) cases. After adjusting for baseline demographic and clinical characteristics, HIV infection was associated with increased mortality but only among those with severe HIV-associated immunosuppression (CD4 count < 200 cells per cubic milliliter), aHR (95% C) = 2.12 (1.23–3.53), p = 0.004, and not among PLWH with ≥ 200 CD4 cells per cubic milliliter, aHR (95% C) = 1.15 (0.82–1.60), p = 0.417.ConclusionAmong relatively young Ugandans, HCC is a devastating disease with rapid mortality that is especially rapid among people living with HIV(PLWH). HIV was associated with slightly higher mortality, notably among PLWH with lower CD4 cell counts. As a substantial majority of PLWH diagnosed with HCC were engaged in HIV care, further investigation should determine the effectiveness of incorporating screening and early identification of HCC among high-risk individuals into existing HIV care programs. Concurrent with growing access to curative localized treatment for HCC in sub-Saharan Africa, leveraging HIV care infrastructure affords opportunities for earlier HCC intervention.

Extramural vascular invasion nomogram before radical resection of rectal cancer based on magnetic resonance imaging.

This study verified the value of magnetic resonance imaging (MRI) to construct a nomogram to preoperatively predict extramural vascular invasion (EMVI) in rectal cancer using MRI characteristics. There were 55 rectal cancer patients with EMVI and 49 without EMVI in the internal training group. The external validation group consisted of 54 rectal cancer patients with EMVI and 55 without EMVI. High-resolution rectal T2WI, pelvic diffusion-weighted imaging (DWI) sequences, and dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) were used. We collected the following data: distance between the lower tumor margin and the anal margin, distance between the lower tumor margin and the anorectal ring, tumor proportion of intestinal wall, mrT stage, maximum tumor diameter, circumferential resection margin, superior rectal vein width, apparent diffusion coefficient (ADC), T2WI EMVI score, DWI and DCE-MRI EMVI scores, demographic information, and preoperative serum tumor marker data. Logistic regression analyses were used to identify independent risk factors of EMVI. A nomogram prediction model was constructed. Receiver operating characteristic curve analysis verified the predictive ability of the nomogram. P < 0.05 was considered significant. Tumor proportion of intestinal wall, superior rectal vein width, T2WI EMVI score, and carbohydrate antigen 19-9 were significant independent predictors of EMVI in rectal cancer and were used to create the model. The areas under the receiver operating characteristic curve, sensitivities, and specificities of the nomogram were 0.746, 65.45%, and 83.67% for the internal training group, respectively, and 0.780, 77.1%, and 71.3% for the external validation group, respectively. A nomogram including MRI characteristics can predict EMVI in rectal cancer preoperatively and provides a valuable reference to formulate individualized treatment plans and predict prognosis.

Toxicity and efficacy of stereotactic body radiation therapy vs. intensity-modulated radiation therapy for the treatment of locally advanced pancreatic cancer in a phase 3 trial.

713 Background: Locally advanced pancreatic cancer (LAPC) remains one of the deadliest cancers. Radiation is listed as a first line therapy option in treatment guidelines and remains a common treatment for this patient population primarily for local disease control and symptoms. Herein, we performed an exploratory analysis to compare the toxicity and efficacy between patients receiving either stereotactic body radiation therapy (SBRT) or intensity-modulated radiation therapy (IMRT) during the induction phase (prior to randomization) of a Phase 3 trial for localized intra-arterial gemcitabine therapy. Methods: As part of the TIGeR-PaC Phase 3 trial, patients with LAPC and an ECOG of 0-1 underwent an induction phase and were treated with 2 cycles of gemcitabine and nab-paclitaxel prior to receiving SBRT (n=59; 33Gy in 5 Fractions) or IMRT (n=75; 50Gy in 25 fractions) with concurrent PO capecitabine BID Monday-Friday. We performed Mann-Whitney analyses to compare mean percent changes in tumor size (imaging performed prior and 1-month after radiation) and CA 19-9 tumor markers. Adverse event (AE) incidence data during and 2 weeks after completion of the last fraction of radiation was used to compare toxicity via a contingency table. Results: A total of 134 patients across 22 sites (63 male/71 female; median age: 68.5 years) underwent radiation with no significant difference in baseline demographics between SBRT and IMRT patients. The decision for SBRT vs. IMRT was site-driven and not pre-specified by the protocol. Of these, 104 had analyzable imaging data, 50 SBRT and 54 IMRT, with a mean baseline tumor size of 4.3 cm and 4.0 cm (p=0.17), respectively. Pancreatic tumor location included head (n=63), body (n=36), or other (n=5). There was no statistical significance between the mean percent change of tumor size between SBRT (-13.3%) and IMRT (-10.8%; p=0.834). Both arms had similar RECIST partial response (approximately 18% of patients) after treatment. Patients with analyzable CA 19-9 tumor marker data (14 SBRT, 17 IMRT) showed no significant difference in percent change (mean, 9.8%±111%, -40.7%±40%; p=0.262). The SBRT subgroup had significantly less patients drop-out due to clinical deterioration (1 SBRT, 9 IMRT; p=0.025) and experience any AE (26 SBRT, 49 IMRT; p=0.014), gastrointestinal AE (10 SBRT, 33 IMRT; p&lt;0.001), grade 3 or higher AE (6 SBRT, 20 IMRT; p=0.017), and serious AE (2 SBRT, 10 IMRT; p=0.045) p=0.014, 0.017, 0.045, respectively). The main AEs and serious AEs were gastrointestinal events. Conclusions: When compared to IMRT, SBRT demonstrates improved tolerability for treatment of patients with LAPC with comparable clinical efficacy. Clinical trial information: NCT03257033 .

Ovarian fibroma in a postmenopausal woman: A case report

Ovarian fibromas are uncommon, accounting for 4% of ovarian tumors. The clinical examination, ultrasonographic, and tumor marker data remain the best preoperative approach currently available for ovarian tumors. However, the diagnosis remains histological. We hereby present the uncommon case of unilateral ovarian fibroma in a 55-year-old woman who presented with pelvic chronic pain the past 2 years. Underwent a bilateral adnexectomy through laparotomy and pathologic examination confirmed the diagnostic of ovarian fibroma. Our case is to illuminate the diagnosis of ovarian fibroma as a rare diagnosis and it should be considered in the differential diagnosis of solid ovarian masses.

Diabetes mellitus in relation to colorectal tumor molecular subtypes: A pooled analysis of more than 9000 cases.

Diabetes is an established risk factor for colorectal cancer. However, colorectal cancer is a heterogeneous disease and it is not well understood whether diabetes is more strongly associated with some tumor molecular subtypes than others. A better understanding of the association between diabetes and colorectal cancer according to molecular subtypes could provide important insights into the biology of this association. We used data on lifestyle and clinical characteristics from the Colorectal Cancer Family Registry (CCFR) and the Genetics and Epidemiology of Colorectal Cancer Consortium (GECCO), including 9756 colorectal cancer cases (with tumor marker data) and 9985 controls, to evaluate associations between reported diabetes and risk of colorectal cancer according to molecular subtypes. Tumor markers included BRAF and KRAS mutations, microsatellite instability and CpG island methylator phenotype. In the multinomial logistic regression model, comparing colorectal cancer cases to cancer‐free controls, diabetes was positively associated with colorectal cancer regardless of subtype. The highest OR estimate was found for BRAF‐mutated colorectal cancer, n = 1086 (ORfully adj: 1.67, 95% confidence intervals [CI]: 1.36‐2.05), with an attenuated association observed between diabetes and colorectal cancer without BRAF‐mutations, n = 7959 (ORfully adj: 1.33, 95% CI: 1.19‐1.48). In the case only analysis, BRAF‐mutation was differentially associated with diabetes (P difference = .03). For the other markers, associations with diabetes were similar across tumor subtypes. In conclusion, our study confirms the established association between diabetes and colorectal cancer risk, and suggests that it particularly increases the risk of BRAF‐mutated tumors.

Long short-term memory model – A deep learning approach for medical data with irregularity in cancer predication with tumor markers

BackgroundMachine learning (ML) has emerged as a superior method for the analysis of large datasets. Application of ML is often hindered by incompleteness of the data which is particularly evident when approaching disease screening data due to varied testing regimens across medical institutions. Here we explored the utility of multiple ML algorithms to predict cancer risk when trained using a large but incomplete real-world dataset of tumor marker (TM) values. MethodsTM screening data were collected from a large asymptomatic cohort (n = 163,174) at two independent medical centers. The cohort included 785 individuals who were subsequently diagnosed with cancer. Data included levels of up to eight TMs, but for most subjects, only a subset of the biomarkers were tested. In some instances, TM values were available at multiple time points, but intervals between tests varied widely. The data were used to train and test various machine learning models to evaluate their robustness for predicting cancer risk. Multiple methods for data imputation were explored and models were developed for both single time-point as well as time-series data. ResultsThe ML algorithm, long short-term memory (LSTM), demonstrated superiority over other models for dealing with irregular medical data. A cancer risk prediction tool was trained and validated for a single time-point test of a TM panel including up to four biomarkers (AUROC = 0.831, 95% CI: 0.827–0.835) which outperformed a single threshold method using the same biomarkers. A second model relying on time series data of up to four time-points for 5 TMs had an AUROC of 0.931. ConclusionsA cancer risk prediction tool was developed by training a LSTM model using a large but incomplete real-world dataset of TM values. The LSTM model was best able to handle irregular data compared to other ML models. The use of time-series TM data can further improve the predictive performance of LSTM models even when the intervals between tests vary widely. These risk prediction tools are useful to direct subjects to further screening sooner, resulting in earlier detection of occult tumors.

Primary ovarian fibroma in a postmenopausal woman: A case report.

IntroductionPrimary ovarian leiomyoma is a rare benign tumor of the ovary seen in women between 20 and 65 years old. Clinical, ultasonographic and tumor marker data remain the best preoperative approach currently available for ovarian tumours. Only pathological examination can establish the diagnosis.Case presentationWe describe a case of unilateral, ovarian leiomyoma. The abdomino-pelvic Computed tomography (CT) and magnetic resonance imaging (MRI) revealed a left adnexial mass. A hysterectomy without adnexal preservation was performed, and histological examination revealed a leiomyoma arising primarily in the ovary. The diagnosis was confirmed immunohistochemically.DiscussionThe tumor may be asymptomatic or may manifest with lower abdominal pain associated to metrorrhagia like in our case. The definitive diagnosis of these lesions is difficult prior to surgical removal. Because there is no pathognomonic symptoms or characteristic imaging findings. The correct diagnosis of an ovarian leiomyoma is confirmed immunohistochemically.ConclusionThis rare tumor of the ovary should be considered in the differential diagnosis of solid ovarian masses. An immunohistochemical analysis is recommended for definitive diagnosis.

PCA, K-PCA, ICA ve Random Forest Kullanarak Yumurtalık Kanserinin Tahmini

Ovarian cancer, which is the most common in women and occurs mostly in the post-menopausal period, develops with the uncontrolled proliferation of the cells in the ovaries and the formation of tumors. Early diagnosis is very difficult and in most cases, it is a type of cancer that is in advanced stages when first diagnosed. While it tends to be treated successfully in the early stages where it is confined to the ovary, it is more difficult to treat in the advanced stages and is often fatal. For this reason, it has been focused on studies that predict whether people have ovarian cancer. In our study, we designed a RF-based ovarian cancer prediction model using a data set consisting of 49 features including blood routine tests, general chemistry tests and tumor marker data of 349 real patients. Since the data set containing too many dimensions will increase the time and resources that need to be spent, we reduced the dimension of the data with PCA, K-PCA and ICA methods and examined its effect on the result and time saving. The best result was obtained with a score of 0.895 F1 by using the new smaller-sized data obtained by the PCA method, in which the dimension was reduced from 49 to 6, in the RF method, and the training of the model took 18.191 seconds. This result was both better as a success and more economical in terms of time spent during model training compared to the prediction made over larger data with 49 features, where no dimension reduction method was used. The study has shown that in predictions made with machine learning models over large-scale medical data, dimension reduction methods will provide advantages in terms of time and resources by improving the prediction results.

Abstract 2619: Defining real-world recurrence in the AACR Project GENIE Biopharma Collaborative Data

Abstract Obtaining information regarding cancer recurrence from a retrospective, EHR-based dataset poses several challenges primarily due to the lack of structured data. Patients are at risk for cancer recurrence beginning at a time point at which they are characterized as having no evidence of disease. The absence of cancer may be indicated on a radiology report or a medical oncologist assessment, requiring manual review and interpretation of potentially ambiguous free text. Further, the recurrence event itself can be defined based on several distinct data sources including pathology, imaging, clinician assessments, or tumor markers. The likelihood of ascertaining recurrence is dependent on the frequency and type of surveillance performed and varies based on tumor type and based on clinicians' thresholds for pursuing workup of borderline or suspicious findings; if follow up assessments are infrequent, there are fewer opportunities to detect recurrence. Given these challenges, there is currently no standardized approach to evaluating cancer recurrence in EHR data, impeding analyses of rare molecular tumor subtypes in multi-institutional linked clinico-genomic databases. For this analysis, we leveraged the AACR Project GENIE Biopharma Collaborative data based on the PRISSMM curation model to develop an algorithm for identifying recurrence among patients diagnosed with stage I-III non-small cell lung cancer or with stage I-III colorectal cancer. This algorithm involves using curated pathology report data to identify a definitive surgery as the time at which patients have completed curative intent treatment. Subsequent imaging reports, pathology reports, medical oncologist assessments and tumor marker data are then evaluated in order to characterize the timing of specific recurrence events. We will present the real-world recurrence algorithm, its underlying rationale and discuss applications of recurrence endpoints. Beyond enabling estimates of recurrence-free survival, identifying cancer recurrence will allow for estimation of progression-free survival among stage I-III patients in addition to estimation of PFS among de novo stage IV patients. Estimating PFS in a large cohort of patients with linked phenomic and genomic data has historically been a limitation of these types of datasets. Overcoming this limitation will allow for precision medicine advances in oncology by facilitating data pooling across institutions and enabling examination of rare molecular subtypes in relation to clinically meaningful endpoints. Citation Format: Jessica A. Lavery, Samantha Brown, Eva Lepisto, Michele L. Lenoue-Newton, Caroline McCarthy, Hira Rizvi, Celeste Yu, Kenneth L. Kehl, Shawn M. Sweeney, Julia E. Rudolph, Nikolaus Schultz, Brooke Mastrogiacomo, Ritika Kundra, Jeremy Warner, Philippe Bedard, Gregory J. Riely, Katherine S. Panageas, Deborah Schrag, AACR Project GENIE Consortium. Defining real-world recurrence in the AACR Project GENIE Biopharma Collaborative Data [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 2619.

Abstract 876: TP53 pathway function, estrogen receptor status, and breast cancer risk factors in the Carolina Breast Cancer Study

Abstract Purpose: TP53 and estrogen receptor (ER) both play essential roles in breast cancer development and progression, with recent research revealing crosstalk between TP53 and ER signaling pathways. While many studies have demonstrated heterogeneity of risk factor associations across ER subtypes, TP53 status has been inconsistently linked to breast cancer risk factors. This may be due to few studies evaluating TP53 effects in the context of ER. Additionally, studies have generally classified TP53 status using immunohistochemistry (IHC) staining or DNA sequencing, which are prone to misclassification of TP53 pathway function. RNA-based methods of measuring TP53 pathway activity may reduce misclassification and clarify etiologic associations. Methods: This case-only analysis included 4,466 incident breast cancer cases from the Carolina Breast Cancer Study (1993-2013). Using RNA expression previously quantified using NanoString assays on FFPE tumor samples, tumors were classified for TP53 functional status (mutant-like or wildtype-like) using a validated 52-gene RNA signature. We used a two-stage polytomous logistic regression model to evaluate sources of risk factor heterogeneity due to RNA-based TP53 or ER, adjusting for each other and for PR, HER2, and grade, while accounting for missing tumor marker data. Risk factor heterogeneity was also evaluated for IHC-based TP53 and ER. For each risk factor, joint effects of TP53 and ER were evaluated by allowing for an interaction between the two markers. Results: When classifying TP53 status using the RNA signature, the effects of several hormone-related factors (oral contraceptive use, menopausal status, age at menopause, and pre- and post-menopausal BMI) were heterogeneous across TP53 subtypes, while heterogeneity of reproductive factors (age at first birth and parity) and smoking status was observed across ER subtypes. The effect of number of births was heterogenous across both TP53 and ER-based subtype definitions. Additionally, we observed an interaction between RNA-based TP53 and ER status with family history of breast cancer in a first-degree relative (p=0.05). When classifying TP53 status using IHC, the TP53 effects were not recapitulated but there was minimal change in the ER effects. We observed an interaction between IHC-based TP53 and ER status with age at menarche (p=0.02), smoking status (p=0.05), and alcohol use (p=0.06). Conclusions: This study demonstrates that TP53 and ER are valuable for defining etiologic subtypes of breast cancer, although IHC measures may not have as much value as RNA. Analyses of the joint effects of TP53 and ER with breast cancer risk factors revealed intriguing findings that could lead to new hypotheses. However, larger studies incorporating multiple correlated tumor characteristics will be required to confirm our findings to conclusively define etiologically relevant subtypes of breast cancer. Citation Format: Amber N. Hurson, Mustapha Abubakar, Kathleen Conway-Dorsey, Katherine Hoadley, Michael I. Love, Andrew F. Olshan, Montserrat Garcia-Closas, Melissa A. Troester. TP53 pathway function, estrogen receptor status, and breast cancer risk factors in the Carolina Breast Cancer Study [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 876.

Impact of pathological response after neoadjuvant chemotherapy on adjuvant therapy decisions and patient outcomes in gastrointestinal cancers.

BackgroundNeoadjuvant chemotherapy (NAC) is frequently used in gastrointestinal cancers (GIC), and pathological, radiological, and tumor marker responses are assessed during and after NAC.AimTo evaluate the relationship between pathologic, radiologic, tumor marker responses and recurrence‐free survival (RFS), overall survival (OS), adjuvant chemotherapy (AC) decisions, and the impact of changing to a different AC regimen after poor response to NAC.Methods and resultsMedical records of GIC patients treated with NAC at Mount Sinai between 1/2012 and 12/2018 were reviewed. One hundred fifty‐six patients (58.3% male, mean age 63 years) were identified. Primary tumor sites were: 43 (27.7%) pancreas, 62 (39.7%) gastroesophageal, and 51 (32.7%) colorectal. After NAC, 31 (19.9%) patients had favorable pathologic response (FPR; defined as College of American Pathologists [CAP] score 0–1). Of 107 patients with radiological data, 59 (55.1%) had an objective response, and of 113 patients with tumor marker data, 61 (54.0%) had a ≥50% reduction post NAC. FPR, but not radiographic or serological responses, was associated with improved RFS (HR 0.28; 95% CI 0.11–0.72) and OS (HR 0.13; 95% CI 0.2–0.94). Changing to a different AC regimen from initial NAC, among all patients and specifically among those with unfavorable pathological response (UPR; defined as CAP score 2–3) after NAC, was not associated with improved RFS or OS.ConclusionsGIC patients with FPR after NAC experienced significant improvements in RFS and OS. Patients with UPR did not benefit from changing AC. Prospective studies to better understand the role of pathological response in AC decisions and outcomes in GIC patients are needed.

The safety and efficacy of PEN-221 somatostatin analog (SSA)-DM1 conjugate in patients (PTS) with advanced GI mid-gut neuroendocrine tumor (NET): Phase 2 results.

4110 Background: PEN-221 is a small molecule drug conjugate composed of a SSTR2 binding somatostatin analog linked to the toxin DM1. PEN-221-001 was a study which assessed the safety, tolerability, pharmacokinetics (PK), and preliminary efficacy of PEN-221 in well differentiated neuroendocrine tumors (NETs) and small cell lung cancer. Here we present the efficacy outcomes for patients enrolled in the GI mid-gut cohort and the safety data for the entire study. Methods: Pts with advanced, SSTR2+ (by imaging) GI mid-gut NETs were enrolled in this cohort of the study. The primary objective was to determine the safety and efficacy of PEN-221 given intravenously, every (q) 3 weeks in patients with documented radiographic progression within the 6 months prior to enrollment. Patients previously treatment with cytotoxic chemotherapy were excluded. Preliminary efficacy was assessed using RECIST 1.1. A clinically meaningful efficacy result was defined as a Clinical Benefit Rate (CBR) &gt; 75% and a median progression-free survival (mPFS) &gt; 8 months. Results: 32 patients (17M/15F) were enrolled between January 2018 to June 2020 and the data cut-off for this report is July 31, 2020. The first nine patients were treated at the phase 1 determined Maximum Tolerated Dose of 18 mg. After review of the safety, tolerability, and PK data from these pts, the regimen was amended to 8.8 mg/m2 for all subsequent pts to achieve more uniform exposures (AUC) across all pts and reduce toxicity in pts with lower body-surface areas (BSA). The mean number of cycles received by pts in this cohort was 7 (range 1-18), with 5 pts still on treatment at time of data lock. PEN-221 was well tolerated in all pts at the dose of 8.8 mg/m2. The most frequent (≥20% pts) PEN-221 related adverse events of any grade were fatigue (39%), nausea (38%), diarrhea (35%), decreased appetite (30%), infusion reaction (24%), AST/ALT/Alk Phos increase (24%), and peripheral neuropathy (21%). Only 14 (10%) of these events were grade 3 or greater. Grade 3 PEN-221 related adverse events which were reported in 2 or more pts included fatigue (7.6%), ALT/AST/Alk Phos increase (7.6%), and peripheral neuropathy (3%). PEN-221 plasma median t1/2 was ̃4.5 h, with exposures uniform using BSA based dosing. Of the 26 pts who were evaluable for response, 23 (88.5%) had stable disease (SD) reported as their best response with a CBR of 88.5%. Target lesion shrinkage was observed in 10 (38%) patients. The median PFS for this cohort was 9 months (CI 5 – 16.5 months). Tumor marker data (Neuron Specific Enolase, Chromogranin A, 5-Hydroxyindoleacetic Acid, and Circulating Tumor Cells) will also be presented. Conclusions: PEN-221 appears well tolerated at 8.8 mg/m2 q 3 weeks and has demonstrated efficacy exceeding its clinical efficacy goals with a CBR of 88.5% and a mPFS of 9 months. A randomized trial of PEN-221 in GI mid-gut NET patients is now in development. Clinical trial information: NCT02936323.

Logistic models and artificial intelligence in the sonographic assessment of adnexal masses - a systematic review of the literature.

Adnexal masses are common, yet challenging, in gynecological practice. Making the differential diagnosis between their benign and malignant condition is essential for optimal surgical management, but reliable pre-surgical differentiation is sometimes difficult using clinical features, ultrasound examination, or tumor markers alone. A possible way to improve the diagnosis is using artificial intelligence (AI) or logistic models developed based on compiling and processing clinical, ultrasound, and tumor marker data together. Ample research has already been conducted in this regard that medical practitioners could benefit from. In this systematic review, we present logistic models and methods using AI, chosen based on their demonstrated high performance in clinical practice. Although some external validation of these models has been performed, further prospective studies are needed in order to select the best model or to create a new, more efficient, one for the pre-surgical evaluation of ovarian masses.

Abstract PO-114: Mathematical modeling of tumor heterogeneity to optimize treatment scheduling and delay the evolution of resistance

Abstract One of the greatest clinical challenges in cancer facing us today is resistance to therapy. In fact, initial response followed by resistance has been observed for the majority of targeted therapies. If a heterogeneous tumor contains even one resistant clone, or the probability of generating a resistance mutation is nonzero, tumor recurrence is theoretically inevitable. New therapies that surpass resistance pathways can take many years—despite extraordinary efforts—before they are available to patients, suggesting that cancers win this arms race against resistance. However, previous literature suggests that it is possible to delay recurrence by optimizing treatment strategies with current therapy options. The purpose of our work is to use computational techniques to predict time to relapse a priori, to identify treatment strategies that could delay therapeutic resistance, and to propose ways these techniques can be applied in a clinical setting. Here, we present a population dynamics model using ordinary differential equations to model tumor growth and evolution over time. Our model simulated the growth dynamics of treatment-sensitive and -resistance subpopulations, allowing the system to incorporate both pre-existing and acquired resistance. We implemented this model into a large-scale in silico experiment with thousands of tumors with different properties, and applied various treatment schedules in order to ascertain the optimal strategy for each simulated tumor. We found that giving short, sharp pulses of therapies followed by breaks in treatment delayed time to relapse in all cases compared to continuous lower-dose treatment. Aggressive tumors were more likely have the highest benefit from pulsed strategies. Furthermore, we found that alternating back-and-forth between two therapies at short, defined intervals led to a later time to relapse than either sequential treatment or combination treatment. These results were robust to stochasticity, suggesting they might apply even in noisy clinical scenarios. To test this hypothesis we applied our model to data from patients with breast cancer. From four patients treated at MSKCC we obtained 1) longitudinal tumor marker data, indicating tumor burden over time, 2) the duration that each treatment was given, and 3) mutational profiles from bulk sequencing, including the cancer cell fraction that harbored each mutation. The first 2-3 data points from each patient were sufficient to tune the model and predict their time to relapse. The model tuned for each patient predicted pulsed treatments that might have delayed the time to relapse. Our results show that non-canonical treatment regimens could be optimized to delay resistance. Models could become more accurate if we sample patients more frequently, with better data on the tumor composition (obtained for example with single-cell sequencing), and if models include the 3-D spatial structure of heterogenous tumors. These features could be important goals for the future of cancer therapy. Citation Format: Deepti Mathur, Bradford Porter Taylor, Walid Chatila, Nikolaus Schultz, Pedram Razavi, Joao Xavier. Mathematical modeling of tumor heterogeneity to optimize treatment scheduling and delay the evolution of resistance [abstract]. In: Proceedings of the AACR Virtual Special Conference on Tumor Heterogeneity: From Single Cells to Clinical Impact; 2020 Sep 17-18. Philadelphia (PA): AACR; Cancer Res 2020;80(21 Suppl):Abstract nr PO-114.

Abstract 620: Evaluating multiple tumor markers in a novel analysis of reproductive factors and breast cancer risk

Abstract Background: Reproductive factors are reported to be differentially associated with risk of breast cancer (BC) subtypes; however, subtype defining markers are highly correlated and clarifying which markers are sources of heterogeneity is challenging. We aimed to define the association of parity, age at first full-term pregnancy (AFFTP), and breastfeeding duration (BFD) with BC subtypes defined by estrogen receptor (ER), progesterone receptor (PR), the human epidermal growth factor receptor 2 (HER2), and grade, and to identify which tumor markers are heterogeneity sources. Methods: We included 56,545 cases and 65,332 controls from 33 population-based studies in the Breast Cancer Association Consortium. A standard polytomous logistic regression with parity (nulliparous (ref), 1, 2, 3, ≥4), AFFTP (&amp;lt;20 years (ref), 20-&amp;lt;25, 25-&amp;lt;30, ≥30), BFD (0 months (ref), &amp;gt;0-6, &amp;gt;6-12, &amp;gt;12-24, &amp;gt;24), age and study was used to estimate odds ratios (ORs) and 95% confidence intervals (CI) with BC subtypes. Subtypes were defined by in situ and invasive intrinsic-like subtypes: luminal A (HR+, HER2-, grade 1&amp;2), luminal B-HER2- (HR+, HER2-, grade 3), luminal B-HER2+ (HR+, HER2+), HER2-enriched (HR-, HER2+) and triple-negative (TN; HR-, HER2-). We used a two-stage polytomous logistic regression to evaluate heterogeneity sources among ER, PR, HER2, grade, and TN specific associations, adjusting for each other and accounting for missing tumor marker data. Results: Luminal A and TN risk patterns were most distinct. Luminal A risk was inversely associated with parity (OR (95%CI): 0.81 (0.73-0.91), 0.71 (0.64-0.79), 0.64 (0.58-0.72), and 0.53 (0.47-0.59), for 1, 2, 3, ≥4 livebirths, respectively), positively associated with later AFTTP (1.00 (0.93-1.08), 1.14 (1.05-1.23), and 1.36 (1.25-1.49) for 20-&amp;lt;25, 25-&amp;lt;30, ≥30 years, respectively) in parous women, but not associated with BFD. In contrast, parous women had a higher TN risk attenuating with more livebirths: 1.38 (1.15-1.65), 1.26 (1.06-1.49), 1.16 (0.97-1.38), and 1.02 (0.84-1.22) for 1, 2, 3, ≥4 livebirths, respectively. TN risk was also inversely associated with late AFFTP (0.92 (0.82-1.03), 0.86 (0.76-0.98), 0.81 (0.69-0.94) for 20-&amp;lt;25, 25-&amp;lt;30, ≥30 years, respectively) as well as with longer BFD (0.93 (0.83-1.04), 0.76 (0.66-0.88), 0.8 (0.68-0.93), and 0.79 (0.64-0.96) for &amp;gt;0-6, &amp;gt;6-12, &amp;gt;12-24, &amp;gt;24 months, respectively). Risk patterns for the other intrinsic-like subtypes and in situ were most like luminal A. Notable sources of heterogeneity for parity and BFD was TN vs non-TN (Phet=0.003 and Phet=0.005, respectively) and for AFFTP were ER (Phet&amp;lt;0.001) and HER2 (Phet=0.003), while grade was not an important source of heterogeneity after accounting for ER, PR, and HER2. Conclusion: We found strong evidence of strong heterogeneity among parity, AFFTP, and BFD with BC subtypes with distinct patterns of associations for TN vs luminal A and other tumor subtypes. Citation Format: Audrey Y. Jung, Thomas U. Ahearn, Sabine Behrens, Haoyu Zhang, Pooja Middha, Marjanka Schmidt, Nilanjan Chatterjee, Montserrat Garcia-Closas, Jenny Chang-Claude, on behalf of the Breast Cancer Association Consortium. Evaluating multiple tumor markers in a novel analysis of reproductive factors and breast cancer risk [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 620.

Abstract 4205: Colorectal cancer molecular markers and subtypes in relation to disease survival

Abstract Colorectal cancer (CRC) is a biologically heterogeneous disease. Such biological heterogeneity reflects diversity in tumor etiology and has implications for survival. Previous studies have demonstrated a strong association between the presence of microsatellite instability (MSI) in CRC and longer disease-specific survival. The influence of other tumor markers, including the CpG island methylator phenotype (CIMP) and BRAF and KRAS mutation status, on survival is less clear. However, these four markers are correlated. Thus, consideration for these markers in combination, with recognition for the distinct etiologies those combinations may reflect, is important to further our understanding of CRC survival. We pooled data from 7 observational studies of CRC: the multisite Colon Cancer Family Registry (CCFR), the Cancer Prevention Study-II (CPS-II), the Diet Activity and Lifestyle Study (DALS), the Health Professionals Follow-Up Study (HPFS), the Nurses' Health Study (NHS), the Melbourne Collaborative Cohort Study (MCCS), and The Cancer Genome Atlas (TCGA). All studies collected data on MSI, CIMP, and somatic mutation status for BRAF and KRAS. Participants with complete tumor marker data were classified into five subtypes based on tumor marker combinations previously proposed to reflect different etiologic pathways: Type 1 (MSI-high, CIMP+, BRAF-mutated, KRAS-wildtype); Type 2 (microsatellite-stable/low-instability (MSS/MSI-low), CIMP+, BRAF-mutated, KRAS-wildtype); Type 3 (MSS/MSI-low, non-CIMP, BRAF-wildtype, KRAS-mutated); Type 4 (MSS/MSI-low, non-CIMP, BRAF-wildtype, KRAS-wildtype); and Type 5 (MSI-high, non-CIMP, BRAF-wildtype, KRAS-wildtype). We used Cox regression to estimate hazard ratios (HR) and 95% confidence intervals (CI) for associations with disease-specific (DSS) and overall survival (OS) for each tumor marker individually (N=4727 to 5105) and for the above tumor marker combinations (N=118 to 2109); analyses were adjusted for age, sex, and study population. When considered individually, MSI-high status was associated with significantly longer (HROS=0.74, HRDSS=0.39, p&amp;lt;0.001) and KRAS-mutated status with significantly shorter survival (HROS=1.11, HRDSS=1.31, p=0.04 and &amp;lt;0.001, respectively); BRAF-mutation and CIMP status were not associated with survival. When markers were combined into the subtypes listed above, relative to those with the predominant Type 4 tumor subtype, participants with Type 2 and 3 CRC had significantly shorter DSS (HRType2=1.73, p=0.002; HRType3=1.28, p=0.002) and those with Type 1 and 5 CRC had longer DSS (HRType1=0.30, p&amp;lt;0.001; HRType5=0.46, p=0.003). Associations were similar for OS. In this large, collaborative study, CRC subtype classified by proposed etiologic pathways were associated with marked differences in survival. These findings highlight the clinical importance of molecular heterogeneity in CRC. Citation Format: Amanda I. Phipps, Peter T. Campbell, Barbara Banbury, Daniel Buchanan, Andrew T. Chan, Dallas English, Alton B. Farris, Graham G. Giles, Tabitha Harrison, Roger Milne, Polly A. Newcomb, Martha L. Slattery, Melissa Southey, Shuji Ogino, Ulrike Peters. Colorectal cancer molecular markers and subtypes in relation to disease survival [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 4205.

Clinical, ultrasound parameters and tumor marker-based mathematical models and scoring systems in pre-surgical diagnosis of adnexal tumors.

The choice of management for patients with adnexal tumors requires careful pre-surgical assessment. In case of adnexal masses, the diagnostic difficulties arise from the heterogenic nature of the adnexal diseases, presence of multiple functional changes, and lack of early symptoms of malignancy. A reliable pre-surgical differentiation cannot be performed using clinical features, ultrasound examination, or tumor markers alone. New diagnostic techniques and novel markers are under investigations, however no single test can be used to conclusively differentiate between malignant and non-malignant adnexal masses. Mathematical models and scoring systems based on different clinical, ultrasonographic and laboratory parameters alone or together may facilitate the diagnosis. Selected mathematical models and scoring systems are presented in this article. Models using only ultrasound features include simple rules, regression models, Gynecologic Imaging Report and Data System, and various morphologic scores. Some logistic regression models are based on multiple clinical and ultrasound data. The OVA1 test is based on five tumor markers without using other data. The Risk of Malignancy Algorithm uses two tumor markers with one clinical parameter. i.e. the menopausal status. Some models used clinical, ultrasound and tumor marker data together. This group of models includes risk of malignancy indices, artificial neural networks, and the ADNEX model. Although some of these models have been compared in the literature, more prospective studies are needed to select the most effective model, to develop the existing models, or to create new more effective models of oncological assessment of the adnexal tumors.

Modification of Occupational Exposures on Bladder Cancer Risk by Common Genetic Polymorphisms.

Few studies have demonstrated gene/environment interactions in cancer research. Using data on high-risk occupations for 2258 case patients and 2410 control patients from two bladder cancer studies, we observed that three of 16 known or candidate bladder cancer susceptibility variants displayed statistically significant and consistent evidence of additive interactions; specifically, the GSTM1 deletion polymorphism (P interaction ≤ .001), rs11892031 (UGT1A, P interaction = .01), and rs798766 (TMEM129-TACC3-FGFR3, P interaction = .03). There was limited evidence for multiplicative interactions. When we examined detailed data on a prevalent occupational exposure associated with increased bladder cancer risk, straight metalworking fluids, we also observed statistically significant additive interaction for rs798766 (TMEM129-TACC3-FGFR3, P interaction = .02), with the interaction more apparent in patients with tumors positive for FGFR3 expression. All statistical tests were two-sided. The interaction we observed for rs798766 (TMEM129-TACC3-FGFR3) with specific exposure to straight metalworking fluids illustrates the value of integrating germline genetic variation, environmental exposures, and tumor marker data to provide insight into the mechanisms of bladder carcinogenesis.