A strategy for full interrogation of prognostic gene expression patterns: exploring the biology of diffuse large B cell lymphoma

Quality-of-life outcomes and risk prediction for patients randomized to nivolumab plus ipilimumab vs nivolumab on LungMAP-S1400I

BACKGROUND

An important issue for patients with cancer treated with novel therapeutics is how they weigh the effects of treatment on survival and quality of life (QOL). We compared QOL in patients enrolled to SWOG S1400I, a substudy of the LungMAP biomarker-driven master protocol.

METHODS

SWOG S1400I was a randomized phase III trial comparing nivolumab plus ipilimumab vs nivolumab for treatment of immunotherapy-naïve disease in advanced squamous cell lung cancer. The primary endpoint was the MD Anderson Symptom Inventory-Lung Cancer severity score at week 7 and week 13 with a target difference of 1.0 points, assessed using multivariable linear regression. A composite risk model for progression-free and overall survival was derived using best-subset selection.

RESULTS

Among 158 evaluable patients, median age was 67.6 years and most were male (66.5%). The adjusted MD Anderson Symptom Inventory-Lung Cancer severity score was 0.04 points (95% confidence interval [CI] = -0.44 to 0.51 points; P = .89) at week 7 and 0.12 points (95% CI = -0.41 to 0.65; P = .66) at week 13. A composite risk model showed that patients with high levels of appetite loss and shortness of breath had a threefold increased risk of progression or death (hazard ratio [HR] = 3.06, 95% CI = 1.88 to 4.98; P < .001) and that those with high levels of both appetite loss and work limitations had a fivefold increased risk of death (HR = 5.60, 95% CI = 3.27 to 9.57; P < .001)-compared with those with neither risk category.

CONCLUSIONS

We found no evidence of a benefit of ipilimumab added to nivolumab compared with nivolumab alone for QOL in S1400I. A risk model identified patients at high risk of poor survival, demonstrating the prognostic relevance of baseline patient-reported outcomes even in those with previously treated advanced cancer.

Predictors of Pain Reduction in Trials of Interventions for Aromatase Inhibitor-Associated Musculoskeletal Symptoms

Background

Almost one-half of aromatase inhibitor (AI)-treated breast cancer patients experience AI-associated musculoskeletal symptoms (AIMSS); 20%-30% discontinue treatment because of severe symptoms. We hypothesized that we could identify predictors of pain reduction in AIMSS intervention trials by combining data from previously conducted trials.

Methods

We pooled patient-level data from 3 randomized trials testing interventions (omega-3 fatty acids, acupuncture, and duloxetine) for AIMSS that had similar eligibility criteria and the same patient-reported outcome measures. Only patients with a baseline Brief Pain Inventory average pain score of at least 4 of 10 were included. The primary outcome examined was 2-point reduction in average pain from baseline to week 12. Variable cut-point selection and logistic regression were used. Risk models were built by summing the number of factors statistically significantly associated with pain reduction. Analyses were stratified by study and adjusted for treatment arm.

Results

For the 583 analyzed patients, the 4 factors statistically significantly associated with pain reduction were Functional Assessment of Cancer Therapy Functional Well-Being greater than 24 and Physical Well-Being greater than 14 (higher scores reflect better function), and Western Ontario and McMaster Universities Osteoarthritis Index less than 50 and Modified Score for the Assessment and Quantification of Chronic Rheumatoid Affections of the Hands less than 33 (lower scores reflect less pain). Patients with all 4 factors were greater than 6 times more likely to experience at least a 2-point pain reduction (odds ratio = 6.37, 95% confidence interval = 2.31 to 17.53, 2-sided P < .001); similar results were found for secondary 30% and 50% pain reduction endpoints.

Conclusions

Patients with AIMSS who have lower symptom and functional distress at study entry on AIMSS intervention trials are more likely to experience meaningful pain reduction. Baseline symptom and functional status should be considered as stratification factors in future interventional trials.

Deregulation of cell adhesion molecules is associated with progression and poor outcomes in endometrial cancer: Analysis of The Cancer Genome Atlas data

Cell adhesion molecules (CAMs) determine the behavior of cancer cells during metastasis. Although some CAMs are dysregulated in certain types of cancer and are associated with cancer progression, to the best of our knowledge, a comprehensive study of CAMs has not been undertaken, particularly in endometrial cancer (EC). In the present study the expression of 225 CAMs in EC patients with various clinicopathological phenotypes were evaluated by statistical analysis using publicly available data from The Cancer Genome Atlas database. The Kaplan-Meier method, and univariate and multivariate Cox proportional hazards regression models were used for survival analyses. Among the differentially expressed CAMs that were associated with aggressive clinicopathological phenotypes, 10 CAM genes were independent prognostic factors compared with other clinicopathological prognostic factors, including stage, grade, age, lymph node status, peritoneal cytology and histological subtype. A total of six genes (L1 cell adhesion molecule, mucin 15, cell surface associated, cell adhesion associated, oncogene regulated, immunoglobulin superfamily member 9B, protocadherin 9 and protocadherin β1) were selected for integrative analysis. The six-gene signature was demonstrated to be an independent prognostic factor and could effectively stratify patients with different risks. Patients with more high-expression CAMs had a higher risk of poor overall survival (OS) rate. The mortality risk for patients with elevation of >4 CAMs was 11 times of that in those without elevation of these 6 CAMs. Similar results were obtained when relapse-free survival (RFS) time was used during the analysis. Prognostic reliability of the six-gene model was validated using data of an independent cohort from the International Cancer Genome Consortium. In conclusion, a combination of CAM alterations contributed to progression and aggressiveness of EC. The six-gene signature was effective for predicting worse OS and RFS in patients with EC and could be complementary to the present clinical prognostic criteria.

Geographic Distribution and Survival Outcomes for Rural Patients With Cancer Treated in Clinical Trials

IMPORTANCE

Studies showing that patients with cancer from rural areas have worse outcomes than their urban counterparts have relied on cancer population data and did not account for differences in access to care. Clinical trial patients receive protocol-directed care by design, so large clinical trial databases are ideal for examining the impact of rural vs urban residency on outcomes.

OBJECTIVE

To compare the geographic distribution and survival outcomes for rural vs urban patients with cancer treated in clinical trials.

DESIGN, SETTING, AND PARTICIPANTS

In this comparative effectiveness retrospective cohort analysis, 36 995 patients from all 50 states enrolled in 44 phase 3 and phase 2/3 SWOG (formerly the Southwest Oncology Group) treatment trials from January 1, 1986, to December 31, 2012, were examined. Seventeen different cancer-specific analysis cohorts were constructed. Data through January 30, 2018, were analyzed.

MAIN OUTCOMES AND MEASURES

Rural vs urban residency was defined using the Rural-Urban Continuum Codes developed by the US Department of Agriculture. Multivariate Cox regression was used to estimate the association of residency with overall survival, progression-free survival, and cancer-specific survival, controlling for major disease-specific prognostic factors and demographic variables and stratifying by study. Different definitions of rurality were examined. The distribution of rural vs urban patients by geographic region was described.

RESULTS

Overall, 27.7% of patients were 65 years or older (range across 17 cohort analyses, 7.8%-74.5%), 40.3% were female in the non-sex-specific analyses (range across 17 cohort analyses, 28.1%-45.9%), and 10.8% were black (range across 17 cohort analyses, 1.9%-22.4%). Overall, 19.4% of patients (7184 of 36 995) were from rural locations. Rural patients were more likely to be aged 65 years or older (rural, 30.7% aged ≥65 years vs urban, 27.0% aged ≥65 years; difference, 3.7%; 95% CI, 2.5%-4.9%; P < .001), were less likely to be black (rural, 5.4% vs urban, 12.1%; difference, 6.7%; 95% CI, 6.1%-7.3%; P < .001), were similar with respect to sex (rural, 40.4% female vs urban, 39.7% female; difference, 0.6%; 95% CI, -1.4% to 2.6%; P = .53), and were well represented within major US geographic regions (West, Midwest, South, and Northeast). Clinical prognostic factors were similar. In multivariable regression, rural patients with adjuvant-stage estrogen receptor-negative and progesterone receptor-negative breast cancer had worse overall survival (hazard ratio, 1.27; 95% CI, 1.06-1.51; P = .008) and cancer-specific survival (hazard ratio, 1.26; 95% CI, 1.04-1.52; P = .02). No other statistically significant differences for overall, progression-free, or cancer-specific survival were found. Results were consistent regardless of the definition of rurality.

CONCLUSIONS AND RELEVANCE

Rural and urban patients with uniform access to cancer care through participation in a SWOG clinical trial had similar outcomes. This finding suggests that improving access to uniform treatment strategies for patients with cancer may help resolve the disparity in cancer outcomes between rural and urban patients.

Novel prognostic genes of diffuse large B-cell lymphoma revealed by survival analysis of gene expression data

Objective

This study aimed to identify prognostic genes for diffuse large B-cell lymphoma (DLBCL), using bioinformatic methods.

Methods

Five gene expression data sets were downloaded from the Gene Expression Omnibus database. Significance analysis of microarrays algorithm was used to identify differentially expressed genes (DEGs) from two data sets. Functional enrichment analysis was performed for the DEGs with the Database for Annotation, Visualization and Integration Discovery (DAVID). Survival analysis was performed with the Kaplan–Meier method using function survfit from package survival of R for the other three data sets. Cox univariate regression analysis was used to further screen out prognostic genes.

Results

Thirty-one common DEGs were identified in the two data sets, mainly enriched in the regulation of lymphocyte activation, immune response, and interleukin-mediated signaling pathway. Combined with 47 DLBCL-related genes acquired by literature retrieval, a total of 78 potential prognostic genes were obtained. Cases from the other three data sets were used in hierarchical clustering, and the 78 genes could cluster them into several subtypes with significant differences in survival curves. Cox univariate regression analysis revealed 45, 33, and eleven prognostic genes in the three data sets, respectively. Five common prognostic genes were revealed, including LCP2, TNFRSF9, FUT8, IRF4, and TLE1, among which LCP2, FUT8, and TLE1 were novel prognostic genes.

Conclusion

Five prognostic genes of DLBCL were identified in this study. They could not only be used for molecular subtyping of DLBCL but also be potential targets for treatment.

Low GILT Expression is Associated with Poor Patient Survival in Diffuse Large B-Cell Lymphoma

The major histocompatibility complex (MHC) class II-restricted antigen processing pathway presents antigenic peptides acquired in the endocytic route for the activation of CD4(+) T cells. Multiple cancers express MHC class II, which may influence the anti-tumor immune response and patient outcome. Low MHC class II expression is associated with poor survival in diffuse large B-cell lymphoma (DLBCL), the most common form of aggressive non-Hodgkin lymphoma. Therefore, we investigated whether gamma-interferon-inducible lysosomal thiol reductase (GILT), an upstream component of the MHC class II-restricted antigen processing pathway that is not regulated by the transcription factor class II transactivator, may be important in DLBCL biology. GILT reduces protein disulfide bonds in the endocytic compartment, exposing additional epitopes for binding to MHC class II and facilitating antigen presentation. In each of four independent gene expression profiling cohorts with a total of 585 DLBCL patients, low GILT expression was significantly associated with poor overall survival. In contrast, low expression of a classical MHC class II gene, HLA-DRA, was associated with poor survival in one of four cohorts. The association of low GILT expression with poor survival was independent of established clinical and molecular prognostic factors, the International Prognostic Index and the cell of origin classification, respectively. Immunohistochemical analysis of GILT expression in 96 DLBCL cases demonstrated variation in GILT protein expression within tumor cells which correlated strongly with GILT mRNA expression. These studies identify a novel association between GILT expression and clinical outcome in lymphoma. Our findings underscore the role of antigen processing in DLBCL and suggest that molecules targeting this pathway warrant investigation as potential therapeutics.

A comparative analysis of prognostic factor models for follicular lymphoma based on a phase III trial of CHOP-rituximab versus CHOP + 131iodine--tositumomab

PURPOSE

There is currently no consensus on optimal frontline therapy for patients with follicular lymphoma. We analyzed a phase III randomized intergroup trial comparing six cycles of CHOP-R (cyclophosphamide-Adriamycin-vincristine-prednisone (Oncovin)-rituximab) with six cycles of CHOP followed by iodine-131 tositumomab radioimmunotherapy (RIT) to assess whether any subsets benefited more from one treatment or the other, and to compare three prognostic models.

EXPERIMENTAL DESIGN

We conducted univariate and multivariate Cox regression analyses of 532 patients enrolled on this trial and compared the prognostic value of the FLIPI (follicular lymphoma international prognostic index), FLIPI2, and LDH + β2M (lactate dehydrogenase + β2-microglobulin) models.

RESULTS

Outcomes were excellent, but not statistically different between the two study arms [5-year progression-free survival (PFS) of 60% with CHOP-R and 66% with CHOP-RIT (P = 0.11); 5-year overall survival (OS) of 92% with CHOP-R and 86% with CHOP-RIT (P = 0.08); overall response rate of 84% for both arms]. The only factor found to potentially predict the impact of treatment was serum β2M; among patients with normal β2M, CHOP-RIT patients had better PFS compared with CHOP-R patients, whereas among patients with high serum β2M, PFS by arm was similar (interaction P value = 0.02).

CONCLUSIONS

All three prognostic models (FLIPI, FLIPI2, and LDH + β2M) predicted both PFS and OS well, though the LDH + β2M model is easiest to apply and identified an especially poor risk subset. In an exploratory analysis using the latter model, there was a statistically significant trend suggesting that low-risk patients had superior observed PFS if treated with CHOP-RIT, whereas high-risk patients had a better PFS with CHOP-R.

Colorimetric in situ hybridization identifies MYC gene signal clusters correlating with increased copy number, mRNA, and protein in diffuse large B-cell lymphoma

Abnormalities of the MYC oncogene on chromosome 8 are characteristic of Burkitt lymphoma and other aggressive B-cell lymphomas, including diffuse large B-cell lymphoma (DLBCL). We recently described a colorimetric in situ hybridization (CISH) method for detecting extra copies of the MYC gene in DLBCL and the frequent occurrence of excess copies of discrete MYC signals in the context of diploidy or polyploidy of chromosome 8, which correlated with increased mRNA signals. We further observed enlarged MYC signals, which were counted as a single gene copy but, by their dimension and unusual shape, likely consisted of "clusters" of MYC genes. In this study, we sought to further characterize these clusters of MYC signals by determining whether the presence of these correlated with other genetic features, mRNA levels, protein, and overall survival. We found that MYC clusters correlated with an abnormal MYC locus and with increased mRNA. MYC mRNA correlated with protein levels, and both increased mRNA and protein correlated with poorer overall survival. MYC clusters were seen in both the germinal center and activated B-cell subtypes of DLBCL. Clusters of MYC signals may be an underappreciated, but clinically important, feature of aggressive B-cell lymphomas with potential prognostic and therapeutic relevance.