Improving diagnostic accuracy of identifying gastric cancer patients with peritoneal metastases: tumor-guided cell-free DNA analysis of peritoneal fluid.

Detection of peritoneal dissemination (PD) in gastric cancer (GC) patients remains challenging. The feasibility of tumor-guided cell-free DNA (cfDNA) detection in prospectively collected peritoneal fluid (ascites and peritoneal lavage) was investigated and compared to conventional cytology in 28 patients. Besides conventional cytology, next generation sequencing was performed on primary tumor DNA and cell-free DNA from peritoneal fluid. Patients were retrospectively grouped into: a positive group (with PD) and a negative group (without PD). Detectable mutations were found in the primary tumor of 68% (n = 19). Sensitivity of PD detection by tumor-guided cfDNA analysis was 91%, compared to 64% by conventional cytology. Within the positive group (n = 11), tumor-guided cfDNA was detected in all patients with ascites samples (4/4, 100%) and in 86% (6/7) of the lavage samples, opposed to 4/4 (100%) patients with ascites and 43% (3/7) with lavage by conventional cytology. Within the negative group (n = 8), conventional cytology was negative for all samples. In two patients, tumor-guided cfDNA was detected in peritoneal lavage fluid. Interestingly, these 2 patients developed PD within 6 months, suggesting a prognostic value of tumor-guided cfDNA detection. This study showed that tumor-guided cfDNA detection in peritoneal fluids of GC patients is feasible and superior to conventional cytology in detecting PD.

Tumor-positive peritoneal cytology in patients with gastric cancer is associated with poor outcome: A nationwide study.

BACKGROUND: The clinical significance of tumor-positive peritoneal cytology (CYT+) in gastric cancer (GC) patients is unclear. This nationwide cohort study aimed to i) assess the frequency of cytological analysis at staging laparoscopy; ii) determine the prevalence of CYT+GC; and iii) compare overall survival (OS) in CYT+ patients versus those with (PM+) and those without (PM-) macroscopic peritoneal disease. METHODS: All patients diagnosed with cT1-4, cN0-2 and M0 or synchronous PM GC between 2016-2021 were identified in the Netherlands Cancer Registry database and linked to the nationwide pathology database. RESULTS: A total of 4397 patients was included, of which 40 % underwent cytological assessment following staging laparoscopy (863/1745). The prevalence of CYT+ was 8 %. A total of 69 patients had CYT+(1.6 %), 789 (17.9 %) had PM+ and 3539 (80.5 %) had PM- disease. Hazard ratio for OS in CYT+ versus PM+ was 0.86 (95 %CI 0.64-1.17, p-value=0.338), and in PM- versus PM+0.43 (95 %CI 0.38-0.49, p-value<0.001). No survival difference was found between systemic chemotherapy versus surgical resection in CYT+ patients. DISCUSSION: In this nationwide study, OS for gastric cancer patients with CYT+ was equally unfavorable as for those with PM+ and significantly worse as compared to those with PM-. The optimal treatment strategy has yet to be established.

Neoadjuvant atezolizumab plus chemotherapy in gastric and gastroesophageal junction adenocarcinoma: the phase 2 PANDA trial.

Gastric and gastroesophageal junction (G/GEJ) cancers carry a poor prognosis, and despite recent advancements, most patients die of their disease. Although immune checkpoint blockade became part of the standard-of-care for patients with metastatic G/GEJ cancers, its efficacy and impact on the tumor microenvironment (TME) in early disease remain largely unknown. We hypothesized higher efficacy of neoadjuvant immunotherapy plus chemotherapy in patients with nonmetastatic G/GEJ cancer. In the phase 2 PANDA trial, patients with previously untreated resectable G/GEJ tumors (n = 21) received neoadjuvant treatment with one cycle of atezolizumab monotherapy followed by four cycles of atezolizumab plus docetaxel, oxaliplatin and capecitabine. Treatment was well tolerated. There were grade 3 immune-related adverse events in two of 20 patients (10%) but no grade 4 or 5 immune-related adverse events, and all patients underwent resection without treatment-related delays, meeting the primary endpoint of safety and feasibility. Tissue was obtained at multiple time points, allowing analysis of the effects of single-agent anti-programmed cell death ligand 1 (PD-L1) and the subsequent combination with chemotherapy on the TME. Twenty of 21 patients underwent surgery and were evaluable for secondary pathologic response and survival endpoints, and 19 were evaluable for exploratory translational analyses. A major pathologic response (≤10% residual viable tumor) was observed in 14 of 20 (70%, 95% confidence interval 46-88%) patients, including 9 (45%, 95% confidence interval 23-68%) pathologic complete responses. At a median follow-up of 47 months, 13 of 14 responders were alive and disease-free, and five of six nonresponders had died as a result of recurrence. Notably, baseline anti-programmed cell death protein 1 (PD-1)+CD8+ T cell infiltration was significantly higher in responders versus nonresponders, and comparison of TME alterations following anti-PD-L1 monotherapy versus the subsequent combination with chemotherapy showed an increased immune activation on single-agent PD-1/L1 axis blockade. On the basis of these data, monotherapy anti-PD-L1 before its combination with chemotherapy warrants further exploration and validation in a larger cohort of patients with nonmetastatic G/GEJ cancer. ClinicalTrials.gov registration: NCT03448835 .

Early Cost-Effectiveness Analysis of Cytoreductive Surgery and Hyperthermic Intraperitoneal Chemotherapy for Gastric Cancer Patients with Limited Peritoneal Carcinomatosis.

BACKGROUND: Gastric cancer patients with peritoneal carcinomatosis (PC) have a poor prognosis, with a median overall survival of 10 months when treated with systemic chemotherapy only. Cohort studies showed that cytoreductive surgery with hyperthermic intraperitoneal chemotherapy (CRS/HIPEC) might improve the prognosis for gastric cancer patients with limited PC. Besides generating trial data on clinical effectiveness, it is crucial to timely collect information on economic aspects to guide the reimbursement decision-making process. No previous data have been published on the cost(-effectiveness) of CRS/HIPEC in this group of patients. Therefore, we performed an early model-based cost-effectiveness analysis of CRS/HIPEC for gastric cancer patients with limited PC in the Dutch setting. METHODS: We constructed a two-state (alive-dead) Markov transition model to evaluate costs and clinical outcomes from a Dutch healthcare perspective. Clinical outcomes, transition probabilities and utilities were derived from literature and verified by clinical experts in the field. Costs were measured using two available representative cohorts (2010-2017): one 'systemic chemotherapy only' cohort and one 'CRS/HIPEC' cohort (n = 10 each). Incremental cost-utility ratios (ICURs) were expressed as Euros per quality-adjusted life-year (QALY). We performed probabilistic and deterministic sensitivity, scenario, and value-of-information analyses using a willingness-to-pay (WTP) threshold of €80,000/QALY, which reflects the Dutch norm for severe diseases. RESULTS: In the base-case analysis, CRS/HIPEC yielded more QALYs (increment of 0.68) and more costs (increment of €34,706) compared with systemic chemotherapy only, resulting in an ICUR of €50,990/QALY. The probability that CRS/HIPEC was cost effective compared with systemic chemotherapy alone was 64%. To reduce uncertainty, the expected value of perfect information amounted to €4,021,468. The scenario analyses did not alter the results and showed that treatment costs, lifetime health-related quality of life and overall survival had the largest influence on the model. CONCLUSIONS: The presented early cost-effectiveness analysis suggests that adding CRS/HIPEC to systemic chemotherapy for gastric cancer patients with limited PC has a good chance of being cost-effectiveness compared with systemic chemotherapy alone when using a WTP of €80,000/QALY. However, there is substantial uncertainty in view of the current available data on effectiveness. Results from the ongoing phase III PERISCOPE II trial are therefore crucial for further decisions on treatment policy and its cost-effectiveness.

IMPemBra: a phase 2 study comparing pembrolizumab with intermittent/short-term dual MAPK pathway inhibition plus pembrolizumab in patients with melanoma harboring the BRAFV600 mutation.

BACKGROUND: Continuous combination of MAPK pathway inhibition (MAPKi) and anti-programmed death-(ligand) 1 (PD-(L)1) showed high response rates, but only limited improvement in progression-free survival (PFS) at the cost of a high frequency of treatment-related adverse events (TRAE) in patients with BRAFV600-mutated melanoma. Short-term MAPKi induces T-cell infiltration in patients and is synergistic with anti-programmed death-1 (PD-1) in a preclinical melanoma mouse model. The aim of this phase 2b trial was to identify an optimal regimen of short-term MAPKi with dabrafenib plus trametinib in combination with pembrolizumab. METHODS: Patients with treatment-naïve BRAFV600E/K-mutant advanced melanoma started pembrolizumab 200 mg every 3 weeks. In week 6, patients were randomized to continue pembrolizumab only (cohort 1), or to receive, in addition, intermittent dabrafenib 150 mg two times per day plus trametinib 2 mg one time per day for two cycles of 1 week (cohort 2), two cycles of 2 weeks (cohort 3), or continuously for 6 weeks (cohort 4). All cohorts continued pembrolizumab for up to 2 years. Primary endpoints were safety and treatment-adherence. Secondary endpoints were objective response rate (ORR) at week 6, 12, 18 and PFS. RESULTS: Between June 2016 and August 2018, 33 patients with advanced melanoma have been included and 32 were randomized. Grade 3-4 TRAE were observed in 12%, 12%, 50%, and 63% of patients in cohort 1, 2, 3, and 4, respectively. All planned targeted therapy was given in 88%, 63%, and 38% of patients in cohort 2, 3, and 4. ORR at week 6, 12, and 18 were 38%, 63%, and 63% in cohort 1; 25%, 63%, and 75% in cohort 2; 25%, 50%, and 75% in cohort 3; and 0%, 63%, and 50% in cohort 4. After a median follow-up of 43.5 months, median PFS was 10.6 months for pembrolizumab monotherapy and not reached for patients treated with pembrolizumab and intermittent dabrafenib and trametinib (p=0.17). The 2-year and 3-year landmark PFS were both 25% for cohort 1, both 63% for cohort 2, 50% and 38% for cohort 3 and 75% and 60% for cohort 4. CONCLUSIONS: The combination of pembrolizumab plus intermittent dabrafenib and trametinib seems more feasible and tolerable than continuous triple therapy. The efficacy is promising and appears to be favorable over pembrolizumab monotherapy. TRIAL REGISTRATION NUMBER: NCT02625337.

The clinical impact of testing for biomarkers in gastric cancer patients: a real-world cohort.

BACKGROUND AND AIMS: In gastric cancer (GC), HER2 was the first biomarker for guided therapy registered for clinical use. Considering the recent approvals of immune check-point blockade (ICB) in gastro-oesophageal cancers, testing for mismatch repair deficiency (dMMR), Epstein-Barr virus (EBV) and PD-L1 combined positive score (CPS) is becoming increasingly important. Here we describe a real-world cohort on biomarker assessment in GC patients. METHODS: Patients diagnosed with GC between 2017 and 2021 were included. Biomarker results were retrieved from electronic patient files. PD-L1 CPS was determined retrospectively on dMMR and EBV-positive (EBV+) tumours. Data on genomic sequencing were analysed separately. RESULTS: Of 363 patients identified, 45% had metastatic disease. In 335 patients (92%) at least one biomarker was tested. The prevalence of HER2+, dMMR and EBV+ tumours was 10% (32 of 319), 7% (20 of 294) and 1% (three of 235), respectively. Of the dMMR and EBV+ tumours, 95% had a PD-L1 CPS ≥ 5. Therapeutic strategy was adjusted in 31 of 55 patients and consisted of anti-HER2 therapies as well as ICB in clinical trials. Genomic alterations were found in 44 of 60 tested patients. TP53 (73%) and PIK3CA (20%) mutations were most common, followed by KRAS mutations (11%) and amplifications (11%). CONCLUSIONS: In this real-world cohort, testing for HER2, dMMR and EBV status affected treatment decisions in 56% of the patients. Although most dMMR and EBV+ tumours had a PD-L1 CPS ≥ 5, not all patients with a high probability of treatment response are identified. Based on these results, a stepwise diagnostic strategy is proposed.

Prognostic value of tumor markers and ctDNA in patients with resectable gastric cancer receiving perioperative treatment: results from the CRITICS trial.

AIM: To evaluate the prognostic value of tumor markers in a European cohort of patients with resectable gastric cancer. METHODS: We performed a post hoc analysis of the CRITICS trial, in which 788 patients received perioperative therapy. Association between survival and pretreatment CEA, CA 19-9, alkaline phosphatase, neutrophils, hemoglobin and lactate dehydrogenase were explored in uni- and multivariable Cox regression analyses. Likelihoods to receive potentially curative surgery were investigated for patients without elevated tumor markers versus one of the tumor markers elevated versus both tumor markers elevated. The association between tumor markers and the presence of circulating tumor DNA (ctDNA) was explored in 50 patients with available ctDNA data. RESULTS: In multivariable analysis, in which we corrected for allocated treatment and other baseline characteristics, elevated pretreatment CEA (HR 1.43; 95% CI 1.11-1.85, p < 0.001) and CA 19-9 (HR 1.79; 95% CI 1.42-2.25, p < 0.001) were associated with worse OS. Likelihoods to receive potentially curative surgery were 86%, 77% and 60% for patients without elevated tumor marker versus either elevated CEA or CA 19-9 versus both elevated, respectively (p < 0.001). Although both preoperative presence of ctDNA and tumor markers were prognostic for survival, no association was found between these two parameters. CONCLUSION: CEA and CA 19-9 were independent prognostic factors for survival in a large cohort of European patients with resectable gastric cancer. No relationship was found between tumor markers and ctDNA. These factors could potentially guide treatment choices and should be included in future trials to determine their definitive position. TRIAL REGISTRATION: ClinicalTrial.gov identifier: NCT00407186. EudraCT number: 2006-00413032.

Towards Personalization in the Curative Treatment of Gastric Cancer.

Gastric cancer is the fifth most common cancer worldwide and has a high mortality rate. In the last decades, treatment strategy has shifted from an exclusive surgical approach to a multidisciplinary strategy. Treatment options for patients with resectable gastric cancer as recommended by different worldwide guidelines, include perioperative chemotherapy, pre- or postoperative chemoradiotherapy and postoperative chemotherapy. Although gastric cancer is a heterogeneous disease with respect to patient-, tumor-, and molecular characteristics, the current standard of care is still according to a one-size-fits-all approach. In this review, we discuss the background of the different treatment strategies in resectable gastric cancer including the current standard, the specific role of radiotherapy, and describe the current areas of research and potential strategies for personalization of therapy.

High XIST and Low 53BP1 Expression Predict Poor Outcome after High-Dose Alkylating Chemotherapy in Patients with a BRCA1-like Breast Cancer.

In previous studies, high expression of XIST and low expression of 53BP1 were respectively associated with poor systemic therapy outcome in patients and therapy resistance in BRCA1-deficient mouse tumor models, but have not been evaluated in BRCA1-deficient patients. Previously, we demonstrated that classifying breast cancer copy number profiles as BRCA1-like or non-BRCA1-like identified patients enriched for defects in BRCA1 that benefit from high-dose (HD) alkylating chemotherapy compared with a conventional standard regimen. We investigated whether XIST and 53BP1 expression predicted poor outcome of HD chemotherapy within 28 BRCA1-like patients from a trial randomizing between HD [4 cycles 5-fluorouracil, epirubicin, cyclophosphamide (FEC) followed by 1 cycle HD carboplatin, thiotepa, cyclophosphamide] or conventional chemotherapy (5 cycles FEC), for which both XIST and 53BP1 statuses were available. High RNA expression of XIST (n = 5) and low protein expression of 53BP1 (n = 3) expression did not coincide. Patients with either one had poor outcome after treatment with HD chemotherapy, whereas patients with low expression of XIST and high expression of 53BP1 derived substantial benefit of this regimen on recurrence-free survival, disease-free survival, and overall survival, corroborating preclinical findings. XIST and 53BP1 may be predictive biomarkers in BRCA1-like breast cancer.

Lack of genomic heterogeneity at high-resolution aCGH between primary breast cancers and their paired lymph node metastases.

Lymph-node metastasis (LNM) predict high recurrence rates in breast cancer patients. Systemic treatment aims to eliminate (micro)metastatic cells. However decisions regarding systemic treatment depend largely on clinical and molecular characteristics of primary tumours. It remains, however, unclear to what extent metastases resemble the cognate primary breast tumours, especially on a genomic level, and as such will be eradicated by the systemic therapy chosen. In this study we used high-resolution aCGH to investigate DNA copy number differences between primary breast cancers and their paired LNMs. To date, no recurrent LNM-specific genomic aberrations have been identified using array comparative genomic hybridization (aCGH) analysis. In our study we employ a high-resolution platform and we stratify on different breast cancer subtypes, both aspects that might have underpowered previously performed studies.To test the possibility that genomic instability in triple-negative breast cancers (TNBCs) might cause increased random and potentially also recurrent copy number aberrations (CNAs) in their LNMs, we studied 10 primary TNBC-LNM pairs and 10 ER-positive (ER+) pairs and verified our findings adding additionally 5 TNBC-LNM and 22 ER+-LNM pairs. We found that all LNMs clustered nearest to their matched tumour except for two cases, of which one was due to the presence of two distinct histological components in one tumour. We found no significantly altered CNAs between tumour and their LNMs in the entire group or in the subgroups. Within the TNBC subgroup, no absolute increase in CNAs was found in the LNMs compared to their primary tumours, suggesting that increased genomic instability does not lead to more CNAs in LNMs. Our findings suggest a high clonal relationship between primary breast tumours and its LNMs, at least prior to treatment, and support the use of primary tumour characteristics to guide adjuvant systemic chemotherapy in breast cancer patients.

Genomic patterns resembling BRCA1- and BRCA2-mutated breast cancers predict benefit of intensified carboplatin-based chemotherapy.

INTRODUCTION: BRCA-mutated breast cancer cells lack the DNA-repair mechanism homologous recombination that is required for error-free DNA double-strand break (DSB) repair. Homologous recombination deficiency (HRD) may cause hypersensitivity to DNA DSB-inducing agents, such as bifunctional alkylating agents and platinum salts. HRD can be caused by BRCA mutations, and by other mechanisms. To identify HRD, studies have focused on triple-negative (TN) breast cancers as these resemble BRCA1-mutated breast cancer closely and might also share this hypersensitivity. However, ways to identify HRD in non-BRCA-mutated, estrogen receptor (ER)-positive breast cancers have remained elusive. The current study provides evidence that genomic patterns resembling BRCA1- or BRCA2-mutated breast cancers can identify breast cancer patients with TN as well as ER-positive, HER2-negative tumors that are sensitive to intensified, DSB-inducing chemotherapy. METHODS: Array comparative genomic hybridization (aCGH) was used to classify breast cancers. Patients with tumors with similar aCGH patterns as BRCA1- and/or BRCA2-mutated breast cancers were defined as having a BRCA-likeCGH status, others as non-BCRA-likeCGH. Stage-III patients (n = 249) had participated in a randomized controlled trial of adjuvant high-dose (HD) cyclophosphamide-thiotepa-carboplatin (CTC) versus 5-fluorouracil-epirubicin-cyclophosphamide (FE90C) chemotherapy. RESULTS: Among patients with BRCA-likeCGH tumors (81/249, 32%), a significant benefit of HD-CTC compared to FE90C was observed regarding overall survival (adjusted hazard ratio 0.19, 95% CI: 0.08 to 0.48) that was not seen for patients with non-BRCA-likeCGH tumors (adjusted hazard ratio 0.90, 95% CI: 0.53 to 1.54) (P = 0.004). Half of all BRCA-likeCGH tumors were ER-positive. CONCLUSIONS: Distinct aCGH patterns differentiated between HER2-negative patients with a markedly improved outcome after adjuvant treatment with an intensified DNA-DSB-inducing regimen (BRCA-likeCGH patients) and those without benefit (non-BRCA-likeCGH patients).

Impact of intertumoral heterogeneity on predicting chemotherapy response of BRCA1-deficient mammary tumors.

The lack of markers to predict chemotherapy responses in patients poses a major handicap in cancer treatment. We searched for gene expression patterns that correlate with docetaxel or cisplatin response in a mouse model for breast cancer associated with BRCA1 deficiency. Array-based expression profiling did not identify a single marker gene predicting docetaxel response, despite an increase in Abcb1 (P-glycoprotein) expression that was sufficient to explain resistance in several poor responders. Intertumoral heterogeneity explained the inability to identify a predictive gene expression signature for docetaxel. To address this problem, we used a novel algorithm designed to detect differential gene expression in a subgroup of the poor responders that could identify tumors with increased Abcb1 transcript levels. In contrast, standard analytical tools, such as significance analysis of microarrays, detected a marker only if it correlated with response in a substantial fraction of tumors. For example, low expression of the Xist gene correlated with cisplatin hypersensitivity in most tumors, and it also predicted long recurrence-free survival of HER2-negative, stage III breast cancer patients treated with intensive platinum-based chemotherapy. Our findings may prove useful for selecting patients with high-risk breast cancer who could benefit from platinum-based therapy.

Genomic instability in breast and ovarian cancers: translation into clinical predictive biomarkers.

Breast and ovarian cancer are among the most common malignancies diagnosed in women worldwide. Together, they account for the majority of cancer-related deaths in women. These cancer types share a number of features, including their association with hereditary cancer syndromes caused by heterozygous germline mutations in BRCA1 or BRCA2. BRCA-associated breast and ovarian cancers are hallmarked by genomic instability and high sensitivity to DNA double-strand break (DSB) inducing agents due to loss of error-free DSB repair via homologous recombination (HR). Recently, poly(ADP-ribose) polymerase inhibitors, a new class of drugs that selectively target HR-deficient tumor cells, have been shown to be highly active in BRCA-associated breast and ovarian cancers. This finding has renewed interest in hallmarks of HR deficiency and the use of other DSB-inducing agents, such as platinum salts or bifunctional alkylators, in breast and ovarian cancer patients. In this review we discuss the similarities between breast and ovarian cancer, the hallmarks of genomic instability in BRCA-mutated and BRCA-like breast and ovarian cancers, and the efforts to search for predictive markers of HR deficiency in order to individualize therapy in breast and ovarian cancer.

Quantitative copy number analysis by Multiplex Ligation-dependent Probe Amplification (MLPA) of BRCA1-associated breast cancer regions identifies BRCAness.

INTRODUCTION: Our group has previously employed array Comparative Genomic Hybridization (aCGH) to assess the genomic patterns of BRCA1-mutated breast cancers. We have shown that the so-called BRCA1-like(aCGH) profile is also present in about half of all triple-negative sporadic breast cancers and is predictive for benefit from intensified alkylating chemotherapy. As aCGH is a rather complex method, we translated the BRCA1(aCGH) profile to a Multiplex Ligation-dependent Probe Amplification (MLPA) assay, to identify both BRCA1-mutated breast cancers and sporadic cases with a BRCA1-like(aCGH) profile. METHODS: The most important genomic regions of the original aCGH based classifier (3q22-27, 5q12-14, 6p23-22, 12p13, 12q21-23, 13q31-34) were mapped to a set of 34 MLPA probes. The training set consisted of 39 BRCA1-like(aCGH) breast cancers and 45 non-BRCA1-like(aCGH) breast cancers, which had previously been analyzed by aCGH. The BRCA1-like(aCGH) group consisted of germline BRCA1-mutated cases and sporadic tumours with low BRCA1 gene expression and/or BRCA1 promoter methylation. We trained a shrunken centroids classifier on the training set and validation was performed on an independent test set of 40 BRCA1-like(aCGH) breast cancers and 32 non-BRCA1-like(aCGH) breast cancer tumours. In addition, we validated the set prospectively on 69 new triple-negative tumours. RESULTS: BRCAness in the training set of 84 tumours could accurately be predicted by prediction analysis of microarrays (PAM) (accuracy 94%). Application of this classifier on the independent validation set correctly predicted BRCA-like status of 62 out of 72 breast tumours (86%). Sensitivity and specificity were 85% and 87%, respectively. When the MLPA-test was subsequently applied to 46 breast tumour samples from a randomized clinical trial, the same survival benefit for BRCA1-like tumours associated with intensified alkylating chemotherapy was shown as was previously reported using the aCGH assay. CONCLUSIONS: Since the MLPA assay can identify BRCA1-deficient breast cancer patients, this method could be applied both for clinical genetic testing and as a predictor of treatment benefit. BRCA1-like tumours are highly sensitive to chemotherapy with DNA damaging agents, and most likely to poly ADP ribose polymerase (PARP)-inhibitors. The MLPA assay is rapid and robust, can easily be multiplexed, and works well with DNA derived from paraffin-embedded tissues.

Ligands of epidermal growth factor receptor and the insulin-like growth factor family as serum biomarkers for response to epidermal growth factor receptor inhibitors in patients with advanced non-small cell lung cancer.

INTRODUCTION: The selection of patients with non-small cell lung cancer (NSCLC) for epidermal growth factor receptor (EGFR) inhibitor (EGFR-tyrosine kinase inhibitors [TKIs]) therapy is suboptimal as tumor tissue is often unavailable. Ligands of EGFR, transforming growth factor-alpha (TGFa) and amphiregulin (ARG), and the insulin-like growth factor (IGF) family have been associated with resistance to EGFR-TKIs. The aim of our study was to explore whether concentrations of these factors measured in serum were predictive of response to EGFR-TKIs. METHODS: We assessed serum levels of marker candidates using enzyme-linked immunosorbent (TGFa and ARG) and chemiluminescent (IGF1 and IGF-binding protein-3) assays in 61 patients with advanced NSCLC treated with EGFR-TKIs and 63 matched advanced NSCLC control patients without EGFR-TKIs treatment. We dichotomized marker levels at the 20th, 50th, or 80th percentile and evaluated whether the effect of EGFR-TKIs treatment on disease-specific survival (DSS) differed by marker level based on multivariate proportional hazards regression with an interaction term. RESULTS: The effect of EGFR-TKIs treatment on DSS showed a significant difference by TGFa and ARG (interaction p = 0.046 and p = 0.004, respectively). Low concentrations of TGFa and high concentrations of ARG were associated with a better DSS in EGFR-TKIs patients compared with control patients. Patients with high concentrations of IGF-binding protein-3 had significantly longer DSS, independent of treatment (hazard ratio: 0.60 per 1 mg/liter, 95% confidence interval: 0.46-0.79). CONCLUSION: Our results suggest that concentrations of TGFa and ARG measured in serum are predictive of EGFR-TKI response. The combination of these two biomarkers could be of value in the process of selecting patients for treatment with EGFR-TKIs.