Virtual Library
Start Your Search
Vera Luiza Capelozzi
Moderator of
-
+
ES 04 - Biology of Lung Cancer (ID 513)
- Event: WCLC 2017
- Type: Educational Session
- Track: Biology/Pathology
- Presentations: 4
- Moderators:Vera Luiza Capelozzi, Akihiko Gemma
- Coordinates: 10/17/2017, 11:00 - 12:30, F201 + F202 (Annex Hall)
-
+
ES 04.01 - Novel Targetable Oncogenes in Lung Cancer (ID 7597)
11:00 - 11:20 | Presenting Author(s): Takashi Kohno
- Abstract
- Presentation
Abstract:
Aberrations of oncogenes, such as EGFR mutation; and ALK and ROS1 fusions, function as a driver in the development of lung adenocarcinoma (LADC) and are established therapeutic targets. We previously identified RET fusion present in 1-2% of LADC (Kohno, Ichikawa, Nat Med, 2012). Its activity as an oncogenic driver in lung carcinogenesis was validated by a study of transgenic mice expressing KIF5B-RET cDNA in their lungs (Saito, Carcinogenesis, 2014). RET fusion is likely to be another target of therapy using tyrosine kinase inhibitors, as represented by a high response rate of RET-fusion positive LADC to vandetanib (LURET study, Yoh et al., Lancet Resp Med, 2016). Our genome-wide sequencing study revealed that ALK, RET and ROS1 oncogene fusion-positive LADCs carry less numbers of mutations in cancer-related genes than others (Saito, Cancer Res, 2015), indicating a small mutation burden in the development of the formers. On the other hand, >30% of LADC and most of other types of lung cancers are negative for the oncogene aberrations above, therefore, other therapeutic targets are needed for precision lung cancer medicine. We have revealed frequent inactivation of chromatin-regulating genes, such as SMARCA4/BRG1 and CREBBP, in lung cancers negative for oncogene aberrations (Ogiwara et al, Cancer Discovery, 2016). We propose a synthetic lethal therapeutic method for chromatin regulator-deficient lung cancers based on inhibition of paralog proteins. LADC driven by somatic EGFR mutations is more prevalent in East Asians (30-50%) than in European/Americans (10-20%). We recently revealed that variations in HLA-class II loci underlie the risk of the disease, by conducting a genome-wide association study of 3,173 EGFR-LADC patients and 15,158 controls (Shiraishi et al., Nature Comm, 2016). The result indicates that LADC develops in vivo through interaction between somatic oncogene mutations and germline variations that modulate immune reaction. We would like to discuss here precision lung cancer medicine based on information on cancer and the host genomes.
Only Members that have purchased this event or have registered via an access code will be able to view this content. To view this presentation, please login, select "Add to Cart" and proceed to checkout. If you would like to become a member of IASLC, please click here.
-
+
ES 04.02 - Resistance Mechanism in TKI (ID 8118)
11:20 - 11:40 | Presenting Author(s): Jürgen Wolf
- Abstract
- Presentation
Abstract:
The implementation of mutation-directed therapy has revolutionized systemic treatment of non-small cell lung cancer (NSCLC). In particular in lung adenocarcinoma targetable driver mutations can be found in a substantial proportion of patients allowing therapy with specific tyrosine kinase inhibitors (TKI) with higher efficacy and better tolerability compared to chemotherapy. Such personalized treatment approaches partly have already become first line standard therapy (EGFR, ALK, ROS1, BRAF V600), other driver mutations are currently evaluated in clinical trials (MET, RET, HER2, NTRK). Also with these new treatment options, however, we are still far away from cure and, mostly after a median progression free survival (PFS) of 10 – 12 months, resistance develops and the patients suffer from relapse. Different mechanisms may underlie primary as well as secondary resistance, which can be subdivided in two major groups: (I) pharmacological resistance, caused by reduced absorption or increased metabolism of the drug or, as a particular challenge in patients with CNS-metastases, by inadequate CNS penetration; (II) biological resistance by molecular changes in the target molecule (resistance mutations or gene copy number gain) or by activation of oncogenic bypass pathways (1). Impressive progress in treating NSCLC patients resistant to TKI therapy has been achieved by a deep understanding of the molecular mechanisms underlying biological resistance, in particular in EGFR mutated and ALK positive NSCLC. In about 60% of patients with acquired resistance (AR) to first- or second generation EGFR-TKIs resistance is caused by the secondary EGFR point mutation T790M leading to reduced TKI binding affinity and conferring growth advantage to the cancer cells. The resistance mechanisms include activation of bypass pathways e.g. by amplification of CMET or HER2 and transition to small cell carcinoma (2). Osimertimib, a third-generation EGFR-TKI, can overcome resistance caused by the T790M mutation and, based on its high clinical activity and favorable tolerability, now has become standard treatment for patients with T790M positive AR to EGFR-TKIs (3). However, resistance also occurs under osimertinib therapy and the molecular mechanisms, which are partly different to those conferring reistance to first generation EGFR inhibitors, are increasingly understood on the molecular level. They include occurrence of the EGFR C797S mutation, activation of the RAS/RAF/MEK/ERK pathway, CMET amplification and HER2 amplification (4,5). Dependent on the molecular mechanism underlying resistance to third generation EGFR inhibitors treatment strategies include the development of next generation inhibitors with activity against C797S positive cancer cells (6) as well as the evaluation of combination therapy approaches e.g. EGFR-TKI plus MET- or MET-inhibitors. These combination approaches are evaluated in the clinical setting of manifest relapse (to overcome resistance) but also as first line treatment (to prevent or postpone relapse). Similarly, also osimertinib is being evaluated in the first line setting and a substantial higher PFS has been reported in this clinical situation. A particular challenge for molecular diagnostics as well as for the development of resistance-overcoming therapeutic strategies is clonal heterogeneity, i.e. the occurrence of different driver mutations within the same patient (7). Also in ALK positive patients substantial progress has been achieved in understanding and overcoming the molecular mechanisms underlying resistance to therapy with ALK-directed TKIs. Similarly to AR to EGFR-TKIs also in AR to ALK-TKIs resistance can be caused by resistance mutations in the ALK receptor itself or by the activation of transforming bypass pathways. A series of resistance mutations has been identified and several next-generation ALK-inhibitors are either already approved or in clinical evaluation. These ALK inhibitors differ in their activity against distinct ALK mutations providing a basis for moleculary guided sequential therapy (8). Already now, impressive prolongation of survival has been reported by the sequential use of the first generation ALK inhibitor crizotinib and the next generation ALK inhibitors alectinib and ceritinib (9,10). Also for other driver mutations like ROS1, BRAF V600 or CMET resistance to TKI therapy is increasingly understood on the molecular level enabling the development of resistance-overcoming treatment strategies for these patients. The development of molecularly guided treatment strategies in AR to TKIs also poses a challenge to molecular diagnostics. In view of the numerous mechanisms which might underly resistance, the implementation of rebiopsies and molecular multiplex diagnostics using next-generation-sequencing (NGS) technologies in clinical routine becomes increasingly important. In addition, the particular challenge of clonal heterogeneity might be addressed already in the near future by the development of highly sensitive NGS-based liquid biopsy diagnostics. References Camidge R et al. Nat Rev Clin Oncol 11, 473-481 (2014) Yu HA et al. Clin Cancer Res 8, 2240-2247 (2013) Mok TS et al. NEJM 7, 629-640 (2017) Thress KS et al. Nat Med 6, 560-562 (2015). Ortiz-Cuaran et al. Clin Cancer Res 19, 4837-4847 (2016) Jia Y et al. Nature 534, 129-132 (2016) Scheffler et al. J Thorac Oncol 10, 40-43 (2015) Gainor et al. Cancer Disc 10, 1118-1133 (2016) Gainor et al. Clin Cancer Res 21, 2745-2752 (2015) Duruisseaux et al. Oncotarget 8, 21903-21017 (2017)
Only Members that have purchased this event or have registered via an access code will be able to view this content. To view this presentation, please login, select "Add to Cart" and proceed to checkout. If you would like to become a member of IASLC, please click here.
-
+
ES 04.03 - Tumor Heterogeneity (ID 7598)
11:40 - 12:00 | Presenting Author(s): Jianjun Zhang
- Abstract
- Presentation
Abstract:
Heterogeneity is a universal phenotype across different cancer types, including non-small cell lung cancer (NSCLC). Tumor heterogeneity is present not only between different tumors from different patients or within the same patients (inter-tumor heterogeneity), but also between different cells within the same tumor (intra-tumor heterogeneity, ITH). ITH results from tumor evolution and in the meanwhile servers as a substrate for tumor evolution. ITH may be influenced by the host antitumor immune surveillance as well as anticancer therapies. Delineating cancer evolution and ITH may provide pivotal insight to our understanding of cancer development, progression and therapeutic resistance, and may eventually help us design more effective preventive and therapeutic strategies. Pioneering studies by multi-region profiling and by comparing paired primary and relapsed tumors have shed light on cancer genomic evolution and suggested the potential impact of genomic ITH on cancer biology and patient outcome. Tracking Non-Small Cell Lung Cancer Evolution through Therapy (TRACERx) is by far the largest study on genomic ITH of NSCLC using multi-region sequencing approach. The results from the first 100 patients enrolled in TRACERx were recently published in the New England Journal of Medicine. In this elegant study, 327 tumor regions from 100 tumors were subjected to high-depth whole exome sequencing (WES). Extensive genomic ITH was demonstrated at both nucleotide and chromosomal levels: a median of 30% of somatic mutations and 48% of copy number alterations (CNAs) were subclonal. Early clonal mutations were associated with smoking signatures in the majority of tumors, while subclonal mutations were significantly enriched for genomic signatures related to spontaneous deamination of methylated cytosines and APOBEC suggesting different mutational mechanisms in play at different times during cancer progression. A high proportion of subclonal CNAs was associated with an increased risk of recurrence and shorter disease free survival (DFS). However, no significant association with DFS was observed between the groups when stratified by proportion of subclonal mutations. The rich data from TRACERx and previous studies are provocative for the future translational and clinical research. Herein, we outline some of the concepts. First, TRACERx provided another piece of evidence that genomic heterogeneity is associated with survival of patients with localized NSCLC. However, it is somewhat surprising that more commonly regarded ITH in point mutations was not found to be associated with survival in this patient cohort, which is in contrast with previous reports in NSCLC and other malignancies. Given the relative small sample size, short postsurgical follow up (median follow up of approximately 18 months with 80% patients less than 2 years) and only 20 relapses, any imbalance in major prognostic factors such as stage, age, histology, smoking, and adjuvant therapy may have masked an actual association between mutational ITH and survival. Our group has recently completed multi-region deep WES on 30 stage IA NSCLC -15 patients relapsed within 3 years post-surgery (cases) and 15 patients have not relapsed with a minimum of 5-year postsurgical follow up (controls). Cases and controls are matched for stage, tumor size, gender, age, histology, smoking history etc. and none of the patients received neoadjuvant or adjuvant therapy. In this well-balanced case-control study, higher degree of point mutation ITH was found to be associated with shorter overall survival and shorter DFS. Nevertheless, the association between CNA ITH and DFS reported in the TRACERx study remained significant after adjusting for known prognostic factors suggests that chromosomal ITH may have greater impact on patient outcome than somatic mutations. This is probably because gain or loss of chromosomal segments or even whole chromosomes could affect hundreds or thousands of genes that may thus disrupt multiple key molecular processes, while point mutations usually affect single genes or pathways. Second, subclonal driver mutations are often detected by multi-region sequencing, which introduces a challenge to our current personalized oncology approach based on sequencing driver genes from single biopsies. Multi-region sequencing is not practical for patients with metastatic diseases or unresectable tumors. However, ctDNA is not spatially limited to certain tumor regions and may have the advantage in detecting subclonal mutations compared to single biopsies. With the rapid progress being made in liquid biopsy and sequencing technologies, sequencing ctDNA could become a practical alternative for multi-region tumor sequencing. Third, majority of studies on NSCLC ITH are based on primarily resected tumors. How chemotherapy, targeted therapy, radiation or immune therapy would impact ITH architectures remains unknown. One can hypothesize that residual tumor cells that survive neoadjuvant therapies could represent the subclones resistant to the these therapies. Therefore, investigating the residual tumors post-neoadjuvant treatment may provide valuable information on mechanisms of drug resistance. As such, well-designed window-of-opportunity neoadjuvant clinical trials would be invaluable for studying drug resistance. Forth, in addition to serving as a potential prognostic biomarker, ITH itself could become a potential therapeutic target. Given the important role of genomic instability in tumor evolution, modulating genomic stability such as targeting APOBEC family, a common cause of subclonal diversification of NSCLC, or inhibiting DNA repair pathways could become a novel therapeutic strategy. This strategy has been recently highlighted by the efficacy of PARP inhibitors in homologous recombination-deficient tumors. Last but not least, the majority of studies on ITH have mainly focused on the genomic ITH. However, ITH can be present at different molecular levels (genetic, epigenetic, gene expression etc.) of cancer cells and also of tumor microenvironment constituting of epithelial cells, blood and lymphatic vessels, cytokines, infiltrating immune cells etc. ITH of any of these components may impact tumor evolution and patient outcome. Our pilot study has demonstrated that a higher level of methylation ITH was associated with larger tumor size, advanced patient age and increased risk of postsurgical recurrence in NSCLC patients. Furthermore, we recently reported substantial T cell repertoire ITH in NSCLC with the majority of T cell clones restricted to individual tumor regions and that a higher degree of T cell repertoire ITH was associated with an increased risk of postsurgical recurrence and shorter DFS. Tumor evolution is a complex process, during which cancer cells accumulate molecular alterations that change their phenotypic features by interacting with the tumor microenvironment. In order to systematically understand the tumor ITH and evolution, future studies are required to depict the overall molecular (genetic, epigenetic, gene and protein expression) ITH of cancer cells as well as the tumor microenvironment components, ideally from longitudinally collected samples with or without treatments to dissect the evolutionary history of NSCLC and other malignancies leading to novel diagnostic, preventive and therapeutic strategies.
Only Members that have purchased this event or have registered via an access code will be able to view this content. To view this presentation, please login, select "Add to Cart" and proceed to checkout. If you would like to become a member of IASLC, please click here.
-
+
ES 04.04 - Exploiting Synthetic Lethality in Lung Cancer Therapy (ID 7866)
12:00 - 12:20 | Presenting Author(s): David P Carbone
- Abstract
- Presentation
Abstract not provided
Only Members that have purchased this event or have registered via an access code will be able to view this content. To view this presentation, please login, select "Add to Cart" and proceed to checkout. If you would like to become a member of IASLC, please click here.
Author of
-
+
MA 01 - SCLC: Research Perspectives (ID 650)
- Event: WCLC 2017
- Type: Mini Oral
- Track: SCLC/Neuroendocrine Tumors
- Presentations: 1
- Moderators:John V Heymach, Eun Kyung Cho
- Coordinates: 10/16/2017, 11:00 - 12:30, Room 503
-
+
MA 01.01 - Metastatic Behavior of Pulmonary Neuroendocrine Carcinomas Is Associated with Epithelial to Mesenchymal Transition Gene Profile (ID 9362)
11:00 - 11:05 | Author(s): Vera Luiza Capelozzi
- Abstract
- Presentation
Background:
The new 2015 WHO classification broadly divided pulmonary neuroendocrine tumor (NET) of the lung in low-grade typical carcinoid (TC) and atypical carcinoid (AC), to the high-grade large-cell neuroendocrine carcinoma (LCNEC) and the small-cell carcinoma (SCLC). The molecular alterations underlying the pathogenesis of these tumors have been studied showing two blocks of entities with independent cellular mechanisms. Many of the differences between the two NETs blocks can be ascribed to tobacco consumption, which induces epithelial to mesenchymal transition activation, responsible for invasive and metastatic behavior. These correlations further highlight the difference in OS for patients with low and high grade metastatic NETs. Therefore, epithelial to mesenchymal transition (EMT) genes profile emerge promise as indicator of invasion and metastasis in NETs.
Method:
Fresh frozen tissue from SCLC (n = 10), LCNEC (n = 4), AC (n = 5), TC (n = 5) and matched normal tissue samples were collected for qRT-PCR analysis carried out on StepOnePlus™ Real-Time PCR System (Applied Biosystems) with RT[2] Profiler PCR Array System for the EMT pathway with 84 target genes (Qiagen, Dusseldorf, Germany). Linear regression was done to evaluate association between gene expressions. Clinical variable such as age, gender, tobacco history, lymph node metastasis and histologic types were associated with gene expression. Differences were regarded as statistically significant at P < 0.05.
Result:
High expression of membrane receptor EGFR (p = 0.003), protein of the matrix metalloproteinase MMP3 (p = 0.044), transcriptional factor TCF3 (p = 0.022) and signaling pathway factor WNT5A (p = 0.013) were observed in patients with tobacco history. Metastatic LCNEC and SCLC presented significant lower expression of JAG1 gene and higher level of EGFR (p<0.01), transmembrane protein DSP (p = 0.03), TCF3 (p = 0.01), TGF-B3 (p = 0.04) and WNT5A (p = 0.01) compared to TC and AC. In addition to these genes, AKT1 and MAP1B were equally high expressed in metastatic NE carcinomas. Importantly, increased expression of these genes added of MMP2 gene was significantly associated with poor OS of the patients.
Conclusion:
A panel of 84 EMT genes was tested and the best biomarkers included EGFR, MMP2, MMP3, TCF3, WNT5A, JAG1, TGFB3, AKT1 and MAP1B with impact on unfavorable prognostic and overall survival of patients, highlight that EMT play a fundamental role in pathogenetic pathway of metastasis in NETs. Supported by CNPq project 301411/2016-6; FAPESP 2013/10113-7.
Only Members that have purchased this event or have registered via an access code will be able to view this content. To view this presentation, please login, select "Add to Cart" and proceed to checkout. If you would like to become a member of IASLC, please click here.
-
+
P2.02 - Biology/Pathology (ID 616)
- Event: WCLC 2017
- Type: Poster Session with Presenters Present
- Track: Biology/Pathology
- Presentations: 1
- Moderators:
- Coordinates: 10/17/2017, 09:30 - 16:00, Exhibit Hall (Hall B + C)
-
+
P2.02-046 - Assessment of PDL1 and Immunoprofiling Using Multiplex Quantitative Immunofluorescence in Lung Cancer: Clinical Implications (ID 10245)
09:30 - 09:30 | Presenting Author(s): Vera Luiza Capelozzi
- Abstract
Background:
Understanding of the “profile” of PD-L1 expression and its interplay with immune cells will provide important insights into lung cancer pathogenesis, and immunotherapeutic strategies targeting this important immune checkpoint protein. The aim was to investigate the correlation between multiplex immunofluorescence (mIF) expression of PD-L1, density and nature of tumor infiltrating immune cells in non-small cell lung carcinomas (NSCLC), and correlate those profiles with clinical and pathological variables including patient outcome.
Method:
We studied 194 stage II/III patients that underwent pulmonary resection, including 98 adenocarcinoma (ADC), 59 squamous cell carcinoma (SqCC), 15 large cells carcinomas (LCC) and 22 neuroendocrine carcinomas (NEC), primary tumors. Formalin-fixed and paraffin embedded (FFPE) tissue microarrays were constructed with five 1.5 mm cores representative of histologic patterns found in each tumor. mIF was performed using the Opal 7-color fIHC Kit™, scanning in the Vectra™ multispectral microscope and analyzed using the inForm™ software (Perkin Elmer, Waltham, MA). The markers studied were grouped in two 6-antibody panels: Panel 1, AE1/AE3 pancytokeratins, PD-L1 (clone E1L3N), PD-1, CD3, CD8 and CD68; and Panel 2, AE1/AE3, Granzyme B, CD45RO and CD57, FOXP3, and CD20. General linear model was used to evaluate the interaction among primary vs metastatic tumors, histologic type and TAICs and Cox's proportional hazard model for overall survival (OS).
Result:
Fifty-eight % out of 164 tumors were positive for PDL-1+ expression (5% cut-off) in malignant cells (EA1/EA3+). Significant higher levels of PD-L1+ expression were detected in NEC compared with other histologies (ADC, SqCC and LCC) (P=0.006). In the same way, we observed higher densities of cytotoxic T lymphocytes (CD3+CD8+) in NEC when compared with the lowest expression in SqCC (P=0.02). Large cell carcinomas presented high levels of memory/regulatory T cells (CD3+FOXP3+CD45RO+) compared with other histologic types but the difference didn´t achieve statistical significance. No difference was found for CD3+PD-L1+, CD68+PD-L1+, natural killer T lymphocytes (CD3+CD57+) and B lymphocytes (CD20+) among the histologic types. Difference between primary and metastatic tumors was found only for naive/memory T lymphocytes (CD3+ CD45RO+) (P=0.04). High CD3+FOXP3+CD45RO+ and CD3+PDL1+ expression were independent favorable prognostic factor for DFS and OS adjusted by smoking, primary vs metastatic, and histologic type [HR 2.68, 95% (CI 1.37–5.24), P=0.004; HR 2.11 (CI 1.07-4.18, P=0.03].
Conclusion:
High abundance of CD3+PD-L1+ cells and memory/regulatory T cells CD3+FOXP3+CD54RO are favorable prognostic factors for resected NSCLC, highlighting the importance of comprehensive assessment of both tumor and immune cells. Supported by CNPq P246042/2012-5 e CNPq 301411/2016-6; FAPESP 2013/10113-7.
-
+
P3.15 - SCLC/Neuroendocrine Tumors (ID 731)
- Event: WCLC 2017
- Type: Poster Session with Presenters Present
- Track: SCLC/Neuroendocrine Tumors
- Presentations: 1
- Moderators:
- Coordinates: 10/18/2017, 09:30 - 16:00, Exhibit Hall (Hall B + C)
-
+
P3.15-006 - Comprehensive Analysis of EMT Gene Signature in Primary and Metastatic Small Cell and Non-Small Cell Carcinomas of the Lung (ID 8182)
09:30 - 09:30 | Author(s): Vera Luiza Capelozzi
- Abstract
Background:
Metastasis are responsible for the death of 90% of patients with lung cancer (LC) indicating the necessity to know the multiple signaling pathways involved. Among them, high-grade neuroendocrine lung carcinomas (NELC) invade and metastasize rapidly. Therefore, biomarkers of aggressiveness in LC remain to be determined, especially in NELC. Epithelial to mesenchymal transition (EMT) genes profile emerge promise as indicator of invasion and metastasis. The aim was to investigate the expression of EMT markers and assessed their relationship with the clinicopathological features and prognosis.
Method:
Fresh frozen tissue from SCLC (n=15) and NSCLC (ADc n=23 and SqCC n=10) and matched normal tissue samples were collected for qRT-PCR analysis carried out on StepOnePlus™ Real-Time PCR with RT[2] Profiler PCR Array System for the EMT pathway with 84 target genes (Qiagen, Dusseldorf, Germany). Gene expression was correlated with clinicopathological variables in the SCLC and NSCLC groups. Survival curves were calculated using the Kaplan-Meier method and risk factors determined by multivariate Cox regression model. Differences were regarded as statistically significant at P<0.05.
Result:
Female patients presented significant higher expression of EGFR (p=0.03), ILK (p=0.05), JAG1 (p=0.01), MMP2 (p=0.04) and SNAI2 (p=0.04) genes. Tobacco history was associated with increased expression of EGFR (p<0.01), ITGAV (p=0.05), SPP1 (p<0.01) and WNT5A (p=0.02). NSCLC presented similar levels of EMT genes evaluated. Tumors from SCLC and NSCLC in advanced N and M stage expressed significant high levels of genes related to cellular membrane [EGFR (p=0.03), ILK (p<0.01), FR11 (p=0.05), ITGAV (p=0.02), ITGB1 (p<0.01), DSP (p=0.04)], extracellular matrix [COL5A2 (p=0.04), COL1A2 (p=0.04)], cytoplasm [GSK3B (p=0.01), VPS13A (p=0.02), MAP1B (p=0.01)] and nucleus SNAI2 (p=0.04). Interestingly, SCLC tumors expressed higher levels of FR11 (p=0.02), GSK3B (p=0.04), ILK (p<0.01), ITGB1 (p=0.01), JAG1 (p<0.01) and MAP1B (p=0.01) indicating more aggressiveness than NSCLC. A mathematical model controlled for N and M stage, histologic type and the gene expression showed that patients with SCLC expressing high levels of MMP2 and SPARC presented significant high risk of death (OR 5.41 and 4.94, respectively) compared to those with lower expression. Patients with NSCLC with low levels of ILK, SPP1, COL1A2, ITGB1 presented a low risk of death (OR -7.02, -0.4, -1.3 and -3.02, respectively).
Conclusion:
Different expression of EMT genes in SCLC and NSCLC, its relationship with histologic types, advanced stage, lymph node metastasis and death suggest a possible role of these markers in their malignancy, but more importantly provide a potential biomolecular marker to predict outcome.
