Determinants of Survival with Combined HER2 and PD-1 Blockade in Metastatic Esophagogastric Cancer

PURPOSE

We report updated clinical outcomes from a phase II study of pembrolizumab, trastuzumab, and chemotherapy (PTC) in metastatic esophagogastric cancer in conjunction with outcomes from an independent Memorial Sloan Kettering (MSK) cohort.

PATIENTS AND METHODS

The significance of pretreatment 89Zr-trastuzumab PET, plasma circulating tumor DNA (ctDNA) dynamics, and tumor HER2 expression and whole exome sequencing was evaluated to identify prognostic biomarkers and mechanisms of resistance in patients treated on-protocol with PTC. Additional prognostic features were evaluated using a multivariable Cox regression model of trastuzumab-treated MSK patients (n = 226). Single-cell RNA sequencing (scRNA-seq) data from MSK and Samsung were evaluated for mechanisms of therapy resistance.

RESULTS

89Zr-trastuzumab PET, scRNA-seq, and serial ctDNA with CT imaging identified how pre-treatment intrapatient genomic heterogeneity contributes to inferior progression-free survival (PFS). We demonstrated that the presence of intensely avid lesions by 89Zr-trastuzumab PET declines in tumor-matched ctDNA by 3 weeks, and clearance of tumor-matched ctDNA by 9 weeks were minimally invasive biomarkers of durable PFS. Paired pre- and on-treatment scRNA-seq identified rapid clearance of HER2-expressing tumor clones with expansion of clones expressing a transcriptional resistance program, which was associated with MT1H, MT1E, MT2A, and MSMB expression. Among trastuzumab-treated patients at MSK, ERBB2 amplification was associated with improved PFS, while alterations in MYC and CDKN2A/B were associated with inferior PFS.

CONCLUSIONS

These findings highlight the clinical relevance of identifying baseline intrapatient heterogeneity and serial ctDNA monitoring of HER2-positive esophagogastric cancer patients to identify early evidence of treatment resistance, which could guide proactive therapy escalation or deescalation.

Prevalence and Severity of Chronic Kidney Disease in Haiti

BACKGROUND

CKD is a major cause of morbidity and mortality in lower-income countries. However, population-based studies characterizing the epidemiology of CKD in these settings are lacking. The study objective was to describe the epidemiology of CKD in a population-based cohort in urban Haiti, including estimates of the prevalence by CKD stage, the magnitude of associated factors with CKD, and the proportion on guideline-recommended treatment.

METHODS

We assessed the prevalence of CKD and associated risk factors in the population-based Haiti Cardiovascular Disease Cohort. We analyzed cross-sectional data from 2424 adults who completed a clinical examination, risk factor surveys, and laboratory measurements for serum creatinine, urinary albumin, and urinary creatinine. We compared our results with US estimates from the National Health and Nutrition Examination Survey. CKD was defined as either a reduced eGFR <60 ml/min per 1.73 m 2 or urinary albumin-to-creatinine ratio ≥30 mg/g according to the Kidney Disease Improving Global Outcomes guidelines. Multivariable logistic regression identified associated factors with CKD.

RESULTS

The mean age was 42 years, 57% of participants were female, and 69% lived in extreme poverty on ≤1 US dollar per day. The age-standardized prevalence of CKD was 14% (95% confidence interval [CI], 12% to 15%). The age-standardized prevalence of reduced eGFR and elevated urinary albumin-to-creatinine ratio was 3% (95% CI, 2% to 4%) and 11% (95% CI, 10% to 13%), respectively. Diabetes (adjusted odds ratio, 4.1; 95% CI, 2.7 to 6.2) and hypertension (adjusted odds ratio, 2.9; 95% CI, 2.0 to 4.2) were significantly associated with CKD. Only 12% of participants with CKD and albuminuria were on guideline-recommended agents, such as angiotensin-converting enzyme inhibitors and angiotensin II receptor blockers.

CONCLUSIONS

In a large population-based cohort of Haitian adults, CKD was highly associated with both diabetes and hypertension. The proportion of participants with CKD on treatment was low, underscoring the need for strengthening clinical management and nephrology care health infrastructure in Haiti.

CLINICAL TRIAL REGISTRY NAME AND REGISTRATION NUMBER

A Longitudinal Cohort Study to Evaluate Cardiovascular Risk Factors and Disease in Haiti, NCT03892265 .

Abstract P559: Evaluating Geographic Variation in Hypertension in Urban Haiti: Findings From the Haiti Cardiovascular Disease Cohort Study

Introduction: Hypertension (HTN) is the leading cardiovascular risk factor in low-income countries including Haiti. Understanding blood pressure (BP) variation at the individual, neighborhood and population levels may help target interventions for multi-level risk reduction. This study aims to identify systolic BP (SBP) and diastolic BP (DBP) variation by neighborhood census blocks in Port-au-Prince to determine if this level should be targeted for future HTN interventions.

Methods: Using data from the Haiti Cardiovascular Disease Cohort (March 2019-August 2021), a population-based cohort of Haitians ≥ 18 years, HTN was defined as SBP ≥140 mmHg or DBP ≥80 mmHg. Census blocks were defined as geographic areas of approximately 100-200 residential buildings per the Haitian Institute of Statistics. The geographic distribution of SBP and DBP was mapped based on household location. The proportion of SBP and DBP variation in block membership was calculated using intra-class correlations (ICC) from linear mixed models.

Results: Among 2,972 participants, median (IQR) age was 40 (28-55) years and 57.6% were female. HTN prevalence was 33.0% and median (IQR) SBP and DBP were 120 (108-140) mmHg and 73.0 (64.0-85.0) mmHg, respectively. No distinct geographic pattern was present for SBP or DBP across blocks (Figure 1). The variation of SBP was 0.7% and DBP was 1.0% across census block membership.

Conclusion: HTN prevalence in urban Port-au-Prince is high without a significant variation in SBP and DBP by census block. This may be due to outdated census data, crossover of individuals living informally across households, and/or the need for different household-level neighborhood geographic constructs such as shared household units. Moreover, these data suggest that interventions may need to target the population level, which span neighborhoods, given the uniformity of hypertension across these communities in Port-au-Prince.

Abstract P130: CVD Risk Factors Among Young People in Haiti: Implications for Low-Income Countries

Introduction: Over 80% of the global cardiovascular disease (CVD) burden is in low-income countries, including Haiti. Emerging data from these settings indicate CVD risk factors occur earlier in the life-course, but our knowledge of the age of onset and associated factors is limited. This study provides the first population-based estimates of CVD risk factors among young people ages 18-30 years in Haiti.

Hypothesis: We hypothesized that elevated blood pressure (pre-hypertension (SBP/DBP >120-140/80-89) or hypertension (SBP/DBP >140/90 or on treatment) is the most common CVD risk factor among young people ages 18-30 years with an estimated prevalence >15%.

Methods: This is a cross-sectional analysis within the Haiti CVD Cohort Study, a population-based longitudinal study of adults ≥18 years in Port au Prince. Participants were enrolled between March 2019-April 2021 using multistage random sampling. At enrollment, sociodemographic and health behaviors were assessed coupled with a physical exam and laboratory measures. CVD risk factors included: elevated BP, dyslipidemia (LDL-C ≥100mg/dL), kidney disease (urine albumin creatinine ratio (ACR) or estimated glomerular filtration rate (eGFR)), obesity (BMI > 30kg/m 2 ), and smoking, alcohol use, and physical activity using the WHO STEPS survey.

Results: Among 957 participants ages 18-30 years, median age was 24 years (IQR 21-27) and 52% were female. Prevalence of elevated BP was 23% (20% pre-hypertension, 3% hypertension), 34% dyslipidemia, 5% ACR ≥30mg/g, 0.21% eGFR < 60 mL/min/1.73, and 7% obese. Other CVD risk factors include: 4% were current smokers, 6% reported any lifetime tobacco use, 19% reported one or more alcoholic drinks per day, and 47% reported low physical activity.

Compared to participants ages 18-24 years, those ages 25-30 years had higher odds of pre-hypertension/hypertension (aOR=1.56, 95% CI: 1.13-2.15), but lower odds of dyslipidemia (aOR=0.55, 95% CI: 0.41-0.74). Females had lower odds of pre-hypertension/hypertension (aOR= 0.24, 95% CI: 0.17-0.35) and dyslipidemia (aOR=0.47, 95% CI: 0.35-0.64). Being obese was associated with a higher odds of pre-hypertension/hypertension (aOR=2.21, 95% CI: 1.14-4.16) but lower odds of dyslipidemia (aOR= 0.35, 95%CI: 0.20-0.61) compared to normal weight.

Conclusion: These data provide previously unknown population-based estimates of CVD risk factors among young people in a low-income country. Rates of elevated BP and dyslipidemia are alarmingly high and occur prematurely in the life-course in this cohort and are associated with risk of future CVD events and mortality. Despite low prevalence of smoking and drinking alcohol, smoking tobacco was still associated with an increased odds of dyslipidemia. Additional drivers of early-onset CVD risk factors must be identified to guide targeted interventions for primordial and primary prevention as well as treatment.

High Dietary Sodium, Measured Using Spot Urine Samples, is Associated with Higher Blood Pressure among Young Adults in Haiti

Background

Hypertension (HTN) is the leading cardiovascular disease (CVD) risk factor in Haiti and is likely driven by poverty-related social and dietary factors. Salt consumption in Haiti is hypothesized to be high but has never been rigorously quantified.

Methods

We used spot urine samples from a subset of participants in the population-based Haiti Cardiovascular Disease Cohort to estimate population mean daily sodium intake. We compared three previously validated formulas for estimating dietary sodium intake using urine sodium, urine creatinine, age, sex, height, and weight. We explored the association between dietary sodium intake and blood pressure, stratified by age group.

Results

A total of 1,240 participants had spot urine samples. Median age was 38 years (range 18-93), and 48% were female. The mean dietary sodium intake was 3.5-5.0 g/day across the three estimation methods, with 94.2%-97.9% of participants consuming above the World Health Organization (WHO) recommended maximum of 2 g/day of sodium. Among young adults aged 18-29, increasing salt intake from the lowest quartile of consumption (<3.73 g/day) to the highest quartile (>5.88 g/day) was associated with a mean 8.71 mmHg higher systolic blood pressure (SBP) (95% confidence interval: 3.35, 14.07; p = 0.001). An association was not seen in older age groups. Among participants under age 40, those with SBP ≥120 mmHg consumed 0.5 g/day more sodium than those with SBP <120 mmHg (95% confidence interval: 0.08, 0.69; p = 0.012).

Conclusions

Nine out of 10 Haitian adults in our study population consumed more than the WHO recommended maximum for daily sodium intake. In young adults, higher sodium consumption was associated with higher SBP. This represents an inflection point for increased HTN risk early in the life course and points to dietary salt intake as a potential modifiable risk factor for primordial and primary CVD prevention in young adults.

Clinical Importance of Clonal Hematopoiesis in Metastatic Gastrointestinal Tract Cancers

Importance

Clonal hematopoiesis (CH) has been associated with development of atherosclerosis and leukemia and worse survival among patients with cancer; however, the association with cancer therapy efficacy, in particular immune checkpoint blockade (ICB), and toxicity has not yet been established. Given the widespread use of ICB and the critical role hematopoietic stem cell-derived lymphocytes play in mediating antitumor responses, CH may be associated with therapeutic efficacy and hematologic toxicity.

Objective

To determine the association between CH and outcomes, hematologic toxicity, and therapeutic efficacy in patients with metastatic gastrointestinal tract cancers being treated with systemic therapy, both in the first-line metastatic treatment setting and in ICB.

Design, Setting, and Participants

This retrospective cohort study included 633 patients with stage IV colorectal (CRC) and esophagogastric (EGC) cancer who were treated with first-line chemotherapy and/or ICB at Memorial Sloan Kettering Cancer Center. Patients underwent matched tumor and peripheral blood DNA sequencing using the Memorial Sloan Kettering-Integrated Mutation Profiling of Actionable Cancer Targets next-generation sequencing assay between January 1, 2006, and December 31, 2020.

Exposures

Clonal hematopoiesis-related genetic alterations were identified by next-generation sequencing of patients' tumor and normal blood buffy coat samples, with a subset of these CH alterations annotated as likely putative drivers (CH-PD) based upon previously established criteria.

Main Outcomes and Measures

Patients with CH and CH-PD in peripheral blood samples were identified, and these findings were correlated with survival outcomes (progression-free survival [PFS] and overall survival [OS]) during first-line chemotherapy and ICB, as well as baseline white blood cell levels and the need for granulocyte colony-stimulating factor (G-CSF) support.

Results

Among the 633 patients included in the study (390 men [61.6%]; median age, 58 [IQR, 48-66] years), the median age was 52 (IQR, 45-63) years in the CRC group and 61 (IQR, 53-69) years in the EGC group. In the CRC group, 161 of 301 patients (53.5%) were men, compared with 229 of 332 patients (69.0%) in the EGC group. Overall, 62 patients (9.8%) were Asian, 45 (7.1%) were Black or African American, 482 (76.1%) were White, and 44 (7.0%) were of unknown race or ethnicity. Presence of CH was identified in 115 patients with EGC (34.6%) and 83 with CRC (27.6%), with approximately half of these patients harboring CH-PD (CRC group, 44 of 83 [53.0%]; EGC group, 55 of 115 [47.8%]). Patients with EGC and CH-PD exhibited a significantly worse median OS of 16.0 (95% CI, 11.6-22.3) months compared with 21.6 (95% CI, 19.6-24.3) months for those without CH-PD (P = .01). For patients with CRC and EGC, CH and CH-PD were not associated with PFS differences in patients undergoing ICB or first-line chemotherapy. Neither CH nor CH-PD were correlated with baseline leukocyte levels or increased need for G-CSF support.

Conclusions and Relevance

These findings suggest CH and CH-PD are not directly associated with the treatment course of patients with metastatic gastrointestinal tract cancer receiving cancer-directed therapy.

Extreme Food Insecurity and Malnutrition in Haiti: Findings from a Population-Based Cohort in Port-au-Prince, Haiti

Haiti is one of the most food-insecure (FIS) nations in the world, with increasing rates of overweight and obesity. This study aimed to characterize FIS among households in urban Haiti and assess the relationship between FIS and body mass index (BMI) using enrollment data from the Haiti Cardiovascular Disease Cohort Study. FIS was characterized as no/low, moderate/high, and extreme based on the Household Food Security Scale. Multinomial logistic generalized estimating equations were used to evaluate the association between FIS categories and BMI, with obesity defined as BMI ≥ 30 kg/m2. Among 2972 participants, the prevalence of moderate/high FIS was 40.1% and extreme FIS was 43.7%. Those with extreme FIS had higher median age (41 vs. 38 years) and were less educated (secondary education: 11.6% vs. 20.3%) compared to those with no/low FIS. Although all FIS categories had high obesity prevalence, those with extreme FIS compared to no/low FIS (15.3% vs. 21.6%) had the lowest prevalence. Multivariable models showed an inverse relationship between FIS and obesity: moderate/high FIS (OR: 0.77, 95% CI: 0.56, 1.08) and extreme FIS (OR: 0.58, 95% CI: 0.42, 0.81) versus no/low FIS were associated with lower adjusted odds of obesity. We found high prevalence of extreme FIS in urban Haiti in a transitioning nutrition setting. The inverse relationship between extreme FIS and obesity needs to be further studied to reduce both FIS and obesity in this population.

High prevalence of obesity among women in urban Haiti: Findings from a population-based cohort

Introduction

Obesity is associated with increased risk of non-communicable diseases and death and is increasing rapidly in low- and middle-income countries, including Haiti. There is limited population-based data on body mass index (BMI) and waist circumference (WC) and associated risk factors in Haiti. This study describes BMI and WC, and factors associated with obesity using a population-based cohort from Port-au-Prince.

Methods

Baseline sociodemographic and clinical data were collected from participants in the Haiti CVD Cohort Study between March 2019 and August 2021. Weight was categorized by BMI (kg/m2) with obesity defined as ≥30 kg/m2. Abdominal obesity was defined using WC cutoffs of ≥80 cm for women and ≥94 cm for men based on WHO guidelines. Sociodemographic and behavioral risk factors, including age, sex, educational attainment, income, smoking status, physical activity, fat/oil use, daily fruit/vegetable consumption, and frequency of fried food intake were assessed for their association with obesity using a Poisson multivariable regression.

Results

Among 2,966 participants, median age was 41 years (IQR: 28-55) and 57.6% were women. Median BMI was 24.0 kg/m2 (IQR: 20.9-28.1) and 508 (17.1%) participants were obese. Women represented 89.2% of the population with BMI ≥30 kg/m2. A total of 1,167 (68.3%) women had WC ≥80 cm and 144 (11.4%) men had WC ≥94 cm. BMI ≥30 kg/m2 was significantly more prevalent among women than men [PR 5.7; 95% CI: (4.3-7.6)], those 40-49 years compared to 18-29 years [PR 3.3; 95% CI: (2.4-4.6)], and those with income >10 USD per day compared to ≤1 USD [PR 1.3; 95% CI: (1.0-1.6)]. There were no significant associations with other health and behavioral risk factors.

Discussion

In Haiti, women have an alarming 6-fold higher obesity prevalence compared to men (26.5 vs. 4.3%) and 89.2% of participants with obesity were women. Abdominal obesity was high, at 44.3%. Haiti faces a paradox of an ongoing national food insecurity crises and a burgeoning obesity epidemic. Individual, social, and environmental drivers of obesity, especially among women, need to be identified.

Polypill for atherosclerotic cardiovascular disease prevention in Haiti: Eligibility estimates in a low-income country

Background

Multidrug therapy is a World Health Organization "best buy" for the prevention and control of noncommunicable diseases. CVD polypills, including ≥2 blood pressure medications, and a statin with or without aspirin, are an effective, scalable strategy to close the treatment gap that exists in many low- and middle-income countries, including Haiti. We estimated the number of Haitian adults eligible for an atherosclerotic CVD (ASCVD) polypill, and the number of potentially preventable CVD events if polypills were implemented nationally.

Methods

We used cross-sectional data from the Haiti CVD Cohort, a population-based cohort of 3,005 adults ≥18 years in Port-au-Prince, to compare two polypill implementation strategies: high-risk primary prevention and secondary prevention. High-risk primary prevention included three scenarios: (a) age ≥40 years, (b) hypertension, or (c) predicted 10-year ASCVD risk ≥7.5%. Secondary prevention eligibility included history of stroke or myocardial infarction. We then used the 2019 Global Burden of Disease database and published polypill trials to estimate preventable CVD events, defined as nonfatal MI, nonfatal stroke, and cardiovascular death over a 5-year timeline.

Results

Among 2,880 participants, the proportion of eligible adults for primary prevention were: 51.6% for age, 32.5% for hypertension, 19.3% for high ASCVD risk, and 5.8% for secondary prevention. Based on current trends, an estimated 462,509 CVD events (95% CI: 369,089-578,475) would occur among adults ≥40 years in Haiti from 2019-2024. Compared with no polypill therapy, we found 32% or 148,003 CVD events (95% CI: 70,126-248,744) could be prevented by a combined primary and secondary prevention approach in Haiti if polypills were fully implemented over 5 years.

Conclusion

These modeling estimates underscore the potential magnitude of preventable CVD events in low-income settings like Haiti. Model calibration using observed CVD events, costs, and implementation assumptions are future directions.

Clinical trial registration

clinicaltrials.gov, identifier: NCT03892265.

Epidermal Growth Factor Receptor Inhibition in Epidermal Growth Factor Receptor-Amplified Gastroesophageal Cancer: Retrospective Global Experience

PURPOSE

Subset analyses from phase III evaluation of epidermal growth factor receptor inhibition (EGFRi) suggest improved outcomes in patients with EGFR-amplified gastroesophageal adenocarcinoma (GEA), but large-scale analyses are lacking. This multi-institutional analysis sought to determine the role of EGFRi in the largest cohort of patients with EGFR-amplified GEA to date.

PATIENTS AND METHODS

A total of 60 patients from 15 tertiary cancer centers in six countries met the inclusion criteria. These criteria required histologically confirmed GEA in the metastatic or unresectable setting with EGFR amplification identified by using a Clinical Laboratory Improvement Amendments-approved assay, and who received on- or off-protocol EGFRi. Testing could be by tissue next-generation sequencing, plasma circulating tumor DNA next-generation sequencing, and/or fluorescence in situ hybridization performed by a Clinical Laboratory Improvement Amendments approved laboratory. Treatment patterns and outcomes analysis was also performed using a deidentified clinicogenomic database (CGDB).

RESULTS

Sixty patients with EGFR-amplified GEA received EGFRi, including 31 of 60 patients (52%) with concurrent chemotherapy. Across treatment lines, patients achieved a 43% objective response rate with a median progression-free survival of 4.6 months (95% CI, 3.5 to 6.4). Patients receiving EGFRi in first-, second-, and third-line therapy achieved a median overall survival of 20.6 months (95% CI, 13.5 to not reached [NR]), 9 months (95% CI, 7.9 to NR), and 8.4 months (7.6 to NR), respectively. This survival far exceeded the 11.2-month (95% CI, 8.7 to 14.2) median overall survival from first-line initiation of non-EGFRi therapy in patients with EGFR-amplified GEA in the CGDB. Despite this benefit, analysis of the CGDB (January 2011-December 2020) suggests that only 5% of patients with EGFR-amplified GEA received EGFRi.

CONCLUSION

Patients with EGFR-amplified GEA derive significant benefit from EGFRi. Further prospective investigation of EGFRi in a well-selected patient population is ongoing in an upcoming trial of amivantamab in EGFR and/or MET amplified GEA.

Clinical relevance of clonal hematopoiesis in metastatic gastrointestinal malignancies

e16082

Background: Clonal hematopoiesis (CH) represents non-random clonal selection of bone marrow-derived cells marked by somatic mutations in certain genes. The presence of CH is associated with development of atherosclerosis and leukemia, and accelerated by toxic exposures (chemotherapy, radiation, smoking) and aging (Jaiswal et al. NEJM 2017; Abelson et al. Nature 2018). The impact of these genetic alterations on cellular function is unknown, especially in the broader context of immunity and in response to cancer therapy. To determine the contribution of CH to therapeutic response and hematologic toxicity in cancer patients, we examined the outcomes of patients treated with cytotoxic and immunotherapy in relationship to CH status. Methods: We evaluated patients with metastatic colorectal cancer (CRC) or esophagogastric cancer (EGC). DNA extracted from whole blood and tumor tissue were sequenced in tandem as part of the MSK-IMPACT hybridization capture-based sequencing assay. CH was defined as any mutation with a VAF of at least 2%, present in at least 10 reads, with at least 2:1 blood:tumor reads, or 1.5:1 blood:lymph node that was not found in gnomAD with a frequency > 0.005. Additional filtering and putative driver definitions (CH-PD) were described by Bolton et al. Nature Genetics 2020. Multivariate survival analyses were performed using a Cox Proportional Hazard model correcting for CH, CH-PD, prior smoking, prior chemotherapy, prior radiation, MSI status, and age at cancer diagnosis. Results: 654 patients with EGC (n = 348) and CRC (n = 306) who began treatment between 2006 and 2020 were included in the analysis. CH was present in 34.5% and 24.4% of each group, and 17.2% and 12.9% harbored CH-PD, respectively. CH and CH-PD were both associated with older age and smoking history, and CH was also associated with prior radiation and MSI-high status (p < 0.05). Patients with CH or CH-PD receiving first-line (1L) therapy for CRC or EGC demonstrated no difference in mPFS after multivariate analysis, though 1L EGC patients with CH-PD had inferior mOS (p = 9e-5). There was no difference in pre-1L WBC, ANC, or ALC, nor in G-CSF or PEG-G-CSF doses administered during 1L therapy between patients with CH or CH-PD versus those without. Similarly, presence of CH or CH-PD had no impact on mPFS or mOS in patients receiving immune checkpoint blockade (ICB) without concurrent chemotherapy after multivariate survival analysis. Conclusions: We confirmed that the mere presence of CH is not prognostic for overall survival, but that EGC patients with CH-PD mutations have inferior overall survival, which is consistent with previous findings. Presence of CH or CH-PD was not associated with differences in baseline leukocyte counts nor need for G-CSF support, nor did it impact PFS in either tumor type, suggesting limited utility of CH in solid tumor clinical decision-making.

Frequency of minimal residual disease as measured by ctDNA in mismatch repair deficient tumors following curative resection

e14520

Background: Mismatch repair deficient (MMRd) tumors are highly sensitive to checkpoint blockade (CPB) in patients with metastatic disease regardless of tumor type. However, the efficacy of CPB in the adjuvant setting is unknown, especially since MMRd is considered a favorable biomarker for most resected tumor types. Circulating tumor DNA (ctDNA) could be used to screen for patients at high risk for recurrence following surgery or adjuvant chemotherapy and identify patients (pts) that would most benefit from CPB. Methods: To assess the frequency of ctDNA in the resected MMRd population, we prospectively screened pts with MMRd tumors who completed standard perioperative chemotherapy and surgery (NCT03832569). DNA from resected tumors and matched postoperative plasma was sequenced for the presence of somatic mutations. Patients were considered to have minimal residual disease (MRD) when mutations were identified in tumor and found to be identical to those in matched plasma DNA. Somatic tissue mutations were assessed using MSK-IMPACT and ctDNA was assessed using FoundationOne, Guardant360 or MSK-ACCESS. Results: A total of 86 pts were screened for the presence of MRD. These represented 7 tumor types with colorectal (63%), endometrial (16%) and esophagogastric (13%) being the most common. The majority of pts were stage III (49%). MRD was detected in 18% of cases (14 of 79). Among the MRD negative group (n=62), only one pt developed disease recurrence. Three samples failed ctDNA analysis for technical reasons. Conclusions: MRD was identified in 18% of resected MMRd tumors using ctDNA analysis, suggesting this to be a feasible tumor agnostic approach to test the efficacy of CPB in a pts at high-risk for recurrence. Future studies will assess the impact of CPB in MRD positive MMRd tumors.[Table: see text]

A nutritional management algorithm in older patients with locally advanced esophageal cancer

e16038

Background: Esophageal cancer primarily affects older adults, who are at highest risk for poor nutritional status due to medical comorbidities, physiological changes of aging and geriatric issues such as altered cognition and mobility. Malnutrition is correlated with poor outcomes in patients with esophageal cancer; however, standardized nutritional interventions are not commonly utilized. Therefore, we performed a feasibility study of a nutritional management algorithm with risk-based guidelines for older patients with esophageal cancer receiving chemoradiation (CRT). Methods: Elderly patients (age ≥ 65 years old) with locally advanced esophageal or gastroesophageal junction (GEJ) cancer receiving induction chemotherapy and preoperative or definitive CRT were eligible for enrollment on this single center study. Patients completed baseline nutritional assessment using the Mini Nutritional Assessment (MNA) screening scale, and patients who were at risk for malnutrition or malnourished were referred to a clinical dietician for evaluation and counseling. Nutritional status was reassessed after induction chemotherapy, and patients with severe malnutrition were to be referred for enteral feeding tube placement prior to CRT. The primary objective was to determine the feasibility of the nutritional management algorithm based on completion rates of nutritional assessment, clinical dietician referral and enteral feeding. Secondary endpoints included toxicity, functional status and quality of life assessment. Results: Twenty elderly patients with locally advanced esophageal cancer were enrolled, and fourteen patients met criteria for clinical dietitian referral based on poor baseline nutritional status. Induction chemotherapy was associated with improved dysphagia, with 92% of patients reporting improvement or resolution of symptoms. There were no patients who met criteria for enteral feeding prior to CRT based on the guidelines, and only one patient (5%) required feeding tube placement during CRT. In total, 17 patients (85%) completed the nutritional management algorithm and finished the planned treatment course. Rates of hospitalization, grade ≥2 esophagitis, grade ≥3 toxicity and early CRT discontinuation were similar between patients with normal and abnormal baseline nutrition. Conclusions: This study demonstrates that a risk-based nutritional management algorithm is feasible in elderly patients with esophageal cancer. The induction chemotherapy approach may ameliorate dysphagia, reduce the need for enteral feeding and facilitate CRT completion in this nutritionally vulnerable population. Clinical trial information: NCT02027948.

Early predictors of benefit to dual anti-PD1/HER2 inhibition: Biomarker analysis from phase 2 trial of pembrolizumab/trastuzumab in HER2-positive metastatic esophagogastric (mEG) cancer

4058

Background: Pembrolizumab and trastuzumab (P&T) and chemotherapy demonstrated 27 month mOS, 13 month mPFS, and 91% response rate in first-line HER2-positive mEG cancer irrespective of PD-L1 status (Janjigian Lancet Oncology 2020). Biomarkers including 89Zr-trastuzumab PET, blood, and tumor analysis were correlated with progression-free survival. Methods: Twenty-five patients received P&T once 3 weeks prior to addition of chemotherapy to P&T. Pre-treatment tumor biopsies, 89Zr-trastuzumab PET scans, serial plasma ctDNA (Guardant360, Redwood City, CA) and CT scans were performed. Tumor-matched DNA alterations were identified by correlating ctDNA and tissue-NGS variant calls. Pre-, on-, and post-treatment biopsies were analyzed using WES and IHC (HER2, PD-L1). Biomarkers were correlated with mPFS and 6-month PFS, the primary endpoint. Results: Of patients with tumor-matched mutations ctDNA at baseline, 12 of 16 had a decline in their maxVAF by week 3, corresponding to a mPFS of 14.7 (11.0-NR) vs 5.9 (95% CI 4.1-NR) months (p=0.009) and a mOS of 29.7 (95% CI 27.2-NR) vs 7.71 (95% CI 6.6-NR) months (p=0.006). 9 of 12 (75%) patients with decline in ctDNA at 3 weeks post-P&T achieved the 6-month PFS primary endpoint while the 4 patients with no decline in ctDNA all progressed in under 6-months. Similarly, 7 of 9 (78%) patients who had a decline in CT-measurements in all disease sites achieved the 6 month PFS primary endpoint, versus 10 of 16 (62.5%) of patients who did not respond in all sites (p=0.66), suggesting that ctDNA is superior to CT as an early predictive biomarker of response. Lack of ERBB2 amplification (amp) by NGS in ctDNA and/or tumor was associated with lack of response to P&T alone prior to addition of chemotherapy. Interestingly, no lesions from patients lacking ERBB2 ctDNA amp (n=3) responded to induction P&T by CT, while lesions from 3/9 patients lacking ERBB2 tissue amp responded to P&T by 3-week CT, suggesting intrapatient HER2 heterogeneity. Eight patients also underwent 89Zr-Trastuzumab PET scans prior to P&T and up to 5 lesions per disease site were measured on serial CT scans. All 15 lesions with intense uptake (SUVmax>10) responded to P&T, but only 9/24 lesions with SUVmax<10. All 4 patients who had at least 1 intense lesion achieved a post-P&T CT response and later 6+ month PFS. All 3 of 3 evaluable patients with intense uptake had baseline ctDNA ERBB2 amp. Conclusions: Patients with a decline in tumor-matched maxVAF after one dose of P&T were more likely to achieve durable PFS. Pre-treatment ctDNA ERBB2 amp and/or intense 89Zr-trastuzumab PET avidity are non-invasive predictive biomarkers of response to HER2-directed therapy. Evaluation of tumor immune environment digital spatial profiling is underway. Clinical trial information: NCT02954536.

Clinical and molecular characteristics of early-onset versus average-onset esophagogastric cancer

250

Background: While the rate of esophagogastric (EG) cancer is declining, early onset (EO) gastric cancer prior to age 50 is rising. It is unknown whether EO-EG cancer represents a distinct entity. This study investigates the clinical and molecular characteristics of EO compared with average onset (AO)-EG cancers. Methods: We reviewed clinical and molecular features of gastric (G), esophageal (E) and gastroesophageal junction (GEJ) cancer in patients treated at MSKCC between 2005 and 2018. We defined early onset as age < 49, based on the age cutoff for urgent endoscopy referral. Clinical symptoms at diagnosis, primary tumor location, histology, HER2 and MSI status and molecular alterations were compared using Fisher’s exact test. Benjamini-Hochberg method was used to decrease the false discovery rate. Results: We analyzed 738 pts with EG cancer (age < 49 n=151; age >50 n=587). Race and sex were different with more Asian (19% vs. 9%), fewer Caucasian (62% vs. 81%) ( P<0.001) and more female patients (40% vs. 29%, P=0.014) in the EO group. Time from symptom onset to diagnosis was longer in the EO group (median (IQR) 144 d (66-276) vs. 75 d (34-136), P=0.009), though stage did not differ ( P=0.49). Patients with EO-EG cancer had less weight loss ( P<0.001), but no other distinct presenting symptoms. Primary disease site was different with more gastric in the EO group (66% vs. 55%, P=0.04). Signet-ring histology was more common in the EO group (11% vs. 3%; P=0.0009). ERBB2 amp and MSI-H were similar, with a trend toward more MSI-H in the AO group (ERBB2 amp P=0.88, Q=0.830; MSI-H P=0.0157, Q=0.056). The most frequent somatic alterations were similar in EO vs. AO pts, including TP53 (68% vs. 70%, P=0.370, Q=0.825), CDH1 (15% vs. 11%, P=0.139, Q=0.825), RHOA (6% vs. 5%, P=0.395, Q=0.825). There was a trend toward more ARID1A (19% vs. 7%, P<0.01, Q=0.250) and FBWX7 (5% vs. 2%, P=139, Q=0.825) mutations in the AO group. Conclusions: Presenting symptoms, stage, histology, HER2 and MSI status are similar in patients with EO vs. AO-EG cancer. There is a trend in EO toward longer time to diagnosis, gastric primary site of disease, signet-ring histology and fewer ARID1A and FBWX7 mutations. Expanded clinical and molecular data will be presented. [Table: see text]

Pembrolizumab with trastuzumab and chemotherapy (PTC) in HER2-positive metastatic esophagogastric cancer (mEG): Plasma and tumor-based biomarker analysis

4559

Background: Pembrolizumab can be safely combined with trastuzumab and chemotherapy and has promising activity with median OS 27 months and 91% objective response rate in HER2-positive mEG cancer irrespective of PD-L1 status (NCT02954536; Janjigian ESMO 2019). Tumor biopsies and blood samples were collected in this phase II trial to identify molecular and immune predictors of response and resistance to PTC. Methods: Pre-treatment and post-progression biopsies were analyzed using WES and IHC (HER2, PD-L1). Peripheral blood was collected pre-treatment, every 9 weeks on-treatment and at progression for plasma ctDNA (Guardant 360, Guardant Health, Redwood, CA). Tumor-matched DNA alterations were identified by correlating ctDNA and solid tumor WES results. Landmark PFS analysis was used to compare ctDNA clearance status at 9 weeks post-treatment. Results: Baseline ctDNA was analysed from 31 of 37 patients of whom 84% (26/31) had tumor-matched ctDNA detected at baseline. Patients who cleared ctDNA at 9 weeks (n = 17/23) achieved a longer median PFS than those who did not - mPFS 12.3 months (95% CI 7.44-NA vs 3.9 months (95% CI 2.01-NA) (log-rank p = 0.02). On serial blood monitoring of 16 patients with eventual radiographic progression, ctDNA re-appearance preceded CT detection in 8 (50%) patients. WES was completed in 31 patients with pre-treatment, and 12 patients post-progression, including matched samples from 10 patients. Loss of HER2 over-expression/amplification was noted in 44% (7/16) of post-progression samples by IHC/FISH (2 IHC 0/1, 5 FISH-). In paired post-progression samples on WES, we observed loss of ERBB2 in 2 patients, and new amplifications of CCND1/3, FGF3/4/19, CDK6/12, KRAS, MYC, and MET, as well as mutations in KRAS, PIK3CD and PIK3RA. Plasma analysis at progression demonstrated copy number increases and/or new amplifications in MET, CKD6, PIK3CA, KRAS, FGFR2, EGFR and CCDN1 as well as KRAS, RB1, PTEN, NF1, NOTCH1, BRAF, and FGFR1 mutations. Conclusions: The majority of patients with previously untreated HER2 positive mEG have detectable plasma ctDNA at baseline. The re-appearance of ctDNA during therapy may serve as an early predictor of progression and help identify genetic drivers of acquired resistance. Loss of ERBB2 over-expression/amplification and activating MAPK alterations occur at PTC progression. Evaluation of tumor immune environment by multiplex IHC and additional ctDNA analysis is underway.

PD-L1 positive esophagogastric (EG) cancer is associated with distinct bacteria

4568

Background: Pembrolizumab is approved for chemotherapy-refractory PD-L1 CPS >1 mEG cancer. In clinical trials, pts with MSI-H, EBV+ and PD-L1 CPS >10 EG cancers derive the greatest benefit with immune checkpoint blockade (ICB). Pre-clinical data suggest that the gut microbiome modulates response to ICB; however, the EG cancer microbiome has not been characterized in EG cancer with respect to PD-L1 and MSI-H status. Therefore, we evaluated the EG tumor microbiome in the context of PD-L1 expression in order to define biologically unique EG tumor phenotypes for future therapeutic development. Methods: Clinical and pathologic characteristics, including age, stage at diagnosis, tumor PD-L1 CPS, HER2 IHC, EBV ISH, genomic analysis, treatment history and survival status were reviewed. CPS was stratified a priori using cutoffs of >1/>10/>20 due to biologic differences. MSK-IMPACT, a capture-based next-generation sequencing platform that detects mutations, copy-number alterations, and select fusions was used to detect non-human bacterial reads identified in the NCBI NT database. Bacterial species found in >2 pts were analyzed and stratified by highest PD-L1 CPS score for each individual patient (Vanderbilt, AMP 2018) and Bonferroni correction was used for odds ratio (OR) confidence intervals where each unique species was considered an independent hypothesis. Results: Molecular data from 311 pts was clinically annotated. PD-L1 results (Table) correlated with bacterial species identified on tumor sequencing. PD-L1 CPS >1 was associated with Selenomonas sputigena (OR: 8.2, 95% CI:1.2-53.6), and PD-L1 CPS >20 was associated with presence of Bifidobacterium dentium (OR: 7.4, 95% CI:1.1-48.5) and Prevotella denticola (OR: 4.2, 95% CI: 1.1-16.6) after multiple comparison correction for the 166 bacterial species identified in the cohort. No differences were seen between PD-L1 < 10 vs >10. Four patients were also found to have EBV+ tumors using this approach, including the 1/54 patients identified by EBER ISH. Conclusions: PD-L1 > 20 EG cancer represents a biologically unique subset, enriched for Bifidobacterium dentium and Prevotella denticola. Correlation between PD-L1 expression, microbial and immune environment, and survival on ICB is underway. [Table: see text]