Classification and Diagnostic Prediction of Cancers using Gene Expression Profiling and Artificial Neural Networks
Javed Khan, MD
Center for Cancer Research, National Cancer Institute
Gene expression profiling, using DNA microarrays, permits a simultaneous analysis of multiple markers, and has been used to categorize cancers into subgroups. However, despite the plethora of statistical techniques to analyze gene expression data, none so far has been rigorously tested for its ability to accurately distinguish cancers belonging to several diagnostic categories.
Artificial neural networks (ANNs) are computer-based algorithms, modeled on the structure and behavior of neurons in the human brain, which can be trained to recognize and categorize complex patterns. Pattern recognition is achieved by adjusting parameters of the ANN by a process of error minimization through learning from experience. They can be calibrated using any type of input data, such as gene expression levels generated by DNA microarrays, and the output can be grouped into any given number of categories. ANNs have been recently applied to clinical problems such as diagnosing myocardial infarcts and arrhythmias from electrocardiograms, and for interpreting radiographs and magnetic resonance images.
We have developed a method of classifying cancers to specific diagnostic categories based on their gene expression signatures using ANN. The ANNs were trained using the small round blue cell tumors (SRBCT) as a model. These cancers belong to four distinct diagnostic categories, and often present diagnostic dilemmas in clinical practice. The ANNs correctly classified all samples and identified the genes most relevant to the classification. Expression of several of these genes has been reported in SRBCT, but the majority have not been previously associated with these cancers. To test the ability of the trained ANN models to recognize SRBCT, we analyzed additional blinded samples that were not previously used for the training procedure, and correctly classified them in all cases. This presentation will demonstrate the potential applications of these methods for tumor diagnosis and the identification of candidate targets for therapy.
Clinical Proteomics: Toward Personalized Molecular Medicine
Lance A. Liotta MD, PhD*+, Elise C. Kohn M.D*., Emanuel F. Petricoin PhD**
*National Cancer Institute, CCR, National Institutes of Health **Center for Biologics Evaluation and Research, FDA
The cause of most human disease lies in the functional disregulation of protein interactions. Understanding the role that protein networks play in disease will create enormous clinical opportunities, since these pathways represent the drug targets of the next decade. In the future, entire cellular networks, not just one disregulated protein, will be the target of therapeutics. The next technologic leap will be the application of proteomic technologies to the bedside. It will soon be possible to analyze the state of protein signal pathways in the disease-altered cells, before, during, and after therapy. This can herald the advent of true patient-tailored therapy.
Ranitidine as adjuvant treatment in colorectal cancer: Influence of blood transfusion and postoperative bacterial infections
Nielsen, Hans Jørgen1; Christensen, Ib J2; Hammer, Janne H1; Moesgaard, Flemming1; Kehlet, Henrik11 Surgical Gastroenterology, Hvidovre University Hospital, Hvidovre, Denmark; 2 Finsen Laboratory, University of Copenhagen Copenhagen, Denmark
Ranitidine may have beneficial effects in patients with colorectal cancer due to angiogenesis inhibition. However, perioperative blood transfusion and postoperative bacterial infections may play significant roles in stimulation of angiogenesis. Therefore, the benefit of ranitidine on long-term survival was evaluated in patients resected for primary colorectal cancer. Patients scheduled for elective resection were consecutively included in a randomized, double blind, placebo controlled, and multicenter clinical study. Just before skin incision ranitidine 100 mg or placebo was given iv. twice daily followed by oral ranitidine 150 mg or placebo twice daily for five years. Adjuvant cytotoxic or radiation therapy was not given. Perioperative blood component transfusion and development of postoperative bacterial infectious complications was prospectively recorded. One month after the operation the patients underwent clinical and biochemical examination, which was repeated every third month until death or withdrawal. Due to the Computerized Personal Registration number given to all Danes none of the patients were lost to follow-up. In an intention-to-treat analysis the effect of ranitidine was evaluated separately in the group of curatively and non-curatively resected patients, and the statistically calculations included perioperative blood transfusion and postoperative bacterial infections. The median observation period was 6.8 (range 5.4-7.9) years of the 740 patients included. In an overall univariate, stratified (at 3.0 years) analysis of 560 curatively resected patients ranitidine improved survival (p=0.04; HR=0.7; 95% CI: 0.5-1.0), while there was no effect on 180 patients with non-curable disease. In separate non-stratified analyses of curatively resected patients, who received perioperative blood transfusion (n=358) ranitidine had no survival benefit, while ranitidine improved survival of 202 non-transfused patients (p=0.02; HR=0.6; 95% CI: 0.4-0.9). However, blood transfusion may increase the frequency of postoperative bacterial infections, and a subsequent analysis of 170 non-transfused patients without development of postoperative bacterial infections showed a significant survival benefit of ranitidine (p=0.01; HR=0.6; 95% CI: 0.4-0.9). In a multivariate analysis, including Dukes’ stage, age, gender, tumor location, blood transfusion, postoperative bacterial infections, and treatment, ranitidine had an independent, beneficial effect on long-term survival (p=0.04; HR=0.6; 95% CI: 0.4-1.0) within the subgroup of patients, who did not receive perioperative blood component transfusion and did not develop postoperative bacterial infections.
Ranitidine may prolong survival of the subgroup of curatively resected colorectal cancer patients, who do not receive perioperative blood transfusion and do not develop postoperative bacterial infections. These two clinical parameters may play detrimental roles for patients treated for a malignant disease.
King, David1; Albertini, Mark1; Schalch, Heidi1; Hank, Jacquelyn1; Surfus, Jean1; Gan, Jacek1; Schiller, Joan H1; Mahvi, David1; Kim, Kyung Mann1; Kendra, Kari1; Reisfeld, Ralph2; Dahl, Thomas3; Kashala, Oscar4; Sturmhoefel, Knut4; Gillies, Steven3; Sondel, Paul11 Univ. of Wisconsin Comprehensive Cancer Center Madison, WI; 2 The Scripps Institute La Jolla, CA; 3 Lexigen Pharmaceuticals Corp. Lexington, MA; 4 EMD Pharmaceuticals Inc. Research Triangle Park, NC
This ongoing Phase I study is evaluating the safety, clinical toxicity and in vivo immunological effects of the hu14.18-IL2 (EMD 273063) immunocytokine (IC) in the treatment of patients with GD2 positive tumors. The hu14.18-IL2 IC consists of the humanized anti-GD2 monoclonal antibody hu14.18, genetically linked to 2 molecules of interleukin-2 (IL-2). The IC targets IL-2 to the tumor microenvironment via the antibody component and directly bridges Fc receptor-bearing effector cells to tumor. Treatment consists of hu14.18-IL2 given as a 4-hour intravenous infusion on Days 1, 2 and 3 of Week 1. Patients with stabilization or regression of disease may receive a second course of treatment at Week 5. As of June 21, 2001, a total of 27 patients with stage IV melanoma (a GD2- positive tumor) have been enrolled. Dose levels of 0.8, 1.6, 3.2, 4.8 and 6.0 mg/m2/day have been evaluated. Fifteen of 25 patients who have completed course 1 had stable disease and went on to complete course 2. Eight patients had stable disease upon completion of 2 courses of treatment. Observed infusion-related Grade 2 adverse events include transient fever, hypotension, chills/rigors, arthralgias, myalgias, nausea, rash, and fatigue. Of the 17 patients who reported arthralgias and/or myalgias, 7 were treated with opioids for relief of symptoms. These myalgias and arthralgias, mainly localized to the lower back and lower extremities, were similar to those seen previously in patients treated with the chimeric form of the 14.18 mAb. There were 7 patients with Grade 3 hypophosphatemia, 2 patients (with known diabetes) with Grade 3 hyperglycemia, 1 patient with Grade 3 hypoxia (observed during marked rigors), 1 patient with Grade 3 thrombocytopenia, 1 patient with a Grade 3 aspartate aminotransferase elevation, and 1 patient with Grade 3 hyperbilirubinemia. No Grade 4 adverse events were observed. Analysis of 24-hr urine phosphates performed on patients with Grade 2 and 3 hypophosphatemia demonstrated no renal wasting of phosphate, thus suggesting migration of extracellular phosphate into the intracellular compartment. Thus transient grade 3 hypophosphatemia was not considered a dose-limiting toxicity. All dose-limiting toxicities resolved spontaneously within a few days. Rebound lymphocytosis, increased numbers of circulating CD16+ and CD56+ cells, and enhanced antibody-dependent cell cytotoxicity by PBMC in vitro were observed, indicating immune activation. Intact hu14.18-IL2 was detected in serum of patients following infusion and appeared to retain activity of the individual antibody and IL-2 components. In conclusion, hu14.18-IL2 has been generally well tolerated at the evaluated dose levels. Toxicities have been those anticipated from prior studies of somewhat similar doses of IL-2 or anti-GD2 mAbs. Immune activation has been demonstrated following hu14.18-IL2 administration. Upon completion of this Phase I study with determination of the MTD, a Phase II clinical treatment protocol is planned to evaluate antitumor activity and in vivo biological-immunological effects of hu14.18-IL2 in patients with advanced melanoma.
Prophylactic Antitumor Efficacy of a Combination Therapy With a Xenogeneic Cell Line Producing Flt3-Ligand (FL),Tumor Cell Lysate, and Interleukin-2 in Mice With Syngeneic Colon Cancer
E.M. Kim, M. Sivanandham, C.I. Stavropoulos, B. Mancke, M.K. Wallack
Department of Surgery, Saint Vincent’s Hospital & Medical Center/New York Medical College, New York, NY
Background: Flt3-Ligand (FL) has been shown to enhance antitumor responses induced by tumor antigen stimulation. Allogeneic and xenogeneic cell lines transduced with cytokine genes have been used to augment the antitumor efficacy of tumor antigens. Objective: To evaluate the augmentation of immunity by a more clinically applicable FL gene-transduced xenogeneic cell line with IL-2 and tumor cell lysate as prophylactic therapy.
Methods: Human 143B Osteosarcoma tumor cells were transduced with full-length FL cDNA (143B-FL). Secretion of FL from 143B-FL was analyzed in vivo in normal BALB/c mice transplanted with 143B-FL, and expansion of dendritic cells (DC) were also analyzed in the same mice by flow cytometry. Eight-week old, male BALB/c mice were used in a prophylactic vaccination protocol. Prophylactic group designations (n=10/group) were as follows: Ten million 143B-FL cells (alone, with tumor lysate, or with tumor lysate and IL-2), IL-2 with tumor lysate or a no treatment control. The tumor lysate (400 ug of protein) and IL-2 (100,000 IU) injections were administered intraperitoneally. Mice were challenged subcutaneously with 103 CC-36 tumor cells. Tumor protection and tumor burden (TB), as mean tumor diameter, were determined. Peripheral Blood Lymphocytes (PBL) from the 143B-FL+IL-2+tumor lysate vaccinated group were analyzed for cytolytic activity in 4-h chromium release assays. In addition, plasma cytokine concentrations of interleukin-12 (IL-12) and interferon gamma (IFN-g) were assessed by ELISA. Student’s t tests were used for all statistical comparisons.
Results: In vivo expression of FL was observed 24 h following the inoculation of 143B-FL, and a 4-fold increase in DCs was observed in the peripheral blood of these mice. Mice immunized with a combination of 143B-FL, tumor lysate and IL-2 showed statistically significant protection against tumor development (10%) for 100 days after tumor challenge; incidences in other groups ranged from 40-100% (p <0.05). Moreover, this immunization protocol produced the lowest TB at 4-and 8- week time points(0,0.2mm) when compared to all other groups (TB between 3.8 – 5.5 mm) (p< 0.05). In addition, PBLs from vaccinated mice showed increased cytolytic activity against CC-36 cells as well as to Yac-1 cells. This corresponded to increased levels of IL-12 and IFN-g in the plasma of mice following vaccination.
Conclusion: These data suggest that vaccination with FL gene-transduced xenogeneic tumor cells with tumor antigens and systemic IL-2 augments immunity through the activation of dendritic cells and T-cell mediated mechanisms.
Inhibition of tumor progression and neoangiogenesis using cyclic RGD-peptides in a chemically induced colon carcinoma in rats
Haier, Joerg1; Goldmann, Ulrike2; Hotz, Birgit2; Keilholz, Ullrich11 Molecular Biology Lab., Dept. of Surgery, University Hospital Muenster Muenster, Germany; 2 Dept. of Surgery, University Hospital B. Franklin, Fu Berlin Berlin, Germany Background:
Integrins are cell surface molecules that mediate cell adhesion, but are also important regulators of tumor cell interactions with their microenvironment, tumor cell survival and growth. In addition, the avb3-integrins appear to be critical for microvessel formation in tumor-induced neoangiogenesis. The present study is the first to investigate the effects of therapeutic avb3-integrin inhibition in a chemically induced tumor model, which largely resembles human colon carcinomas. Methods: Tumor induction was performed in 47 male Sprague-Dawley rats using 1,2 Dimethylhydrazin (21 mg/kg) twice a week. After 20 weeks of tumor induction 100% of the animals developed adenocarcinomas with a median of 13.5 macroscopic tumor nodules (range 12-17), but no distant metastases. During further tumor induction for additional 10 weeks rats were treated three times/week with (a) 15 mg/kg RGDfV-peptide that can block vitronectin and fibronectin receptors; (b) an equimolar amount of an ineffective cyclic control peptide; or (c) with equimolar amounts of a linear RGDS-peptide. At the end of this treatment period rats were sacrificed, and tumor load was quantified macroscopically and confirmed by histological examination. For investigation of the involvement of tumor-induced neoangiogenesis microvessel density was determined using CD31-immunostaining. Results: After 30 weeks control animals (group B) had 5-18 tumors (median 14.5). If rats were treated with RGDfV-peptide (group A) the numbers of tumor nodules was significantly reduced (p<0.005) to a median of 7 macroscopic tumors (range 2-10 tumors) which also represented a significant reduction (p<0.005) compared to prior to treatment. Application of noncylic RGDS-peptides (group C) did not affect the numbers of tumor nodules (median 18; range 10-30 tumors). The diameters of tumor nodules were comparable (3.2-6.1 mm) in animals of all groups. In addition, microvessel density was significantly (p<0.05) reduced in tumors in group A compared to control rats. The major side effect in the treatment group was increased susceptibility to respiratory infections. Conclusions: Our results demonstrate that avb3-integrin-receptor inhibition appears to be a therapeutic strategy for colorectal cancer. In our therapeutic model late onset of treatment with integrin-blocking peptides resulted in an inhibition of tumor growth and a reduced tumor load which appears to be mediated, at least in part, by inhibition of neoangiogenesis.