An Automated and Quantitative Protein Expression-Based Classification System to Identify High Risk Multiple Myeloma Patients
Bibliographic record
Abstract
Abstract Background: Autologous stem cell transplantation (ASCT) has dramatically improved the survival of myeloma patients; however, this approach has significant toxicities and nearly 25% of MM patients progress within one year from their transplant. While gene expression profiling-based (GEP) molecular classification has permitted the identification of unresponsive high-risk patients, these approaches have proven too costly and complex to translate into clinical practice. Less expensive and more readily available methods are needed clinically to identify, at the time of diagnosis, MM patients who may benefit from more aggressive or experimental therapies. While protein-based tissue arrays offer such alternative, biases introduced by the “observer-dependent” scoring methods have limited their wide applicability. Methods: We have designed a simplified, fully automated and quantitative protein expression based-classification system that will allow us to accurately predict survival post ASCT in a cost effective and “observer-independent” manner. We constructed tissue microarrays using diagnostic bone marrow biopsies of 82 newly diagnosed MM patients uniformly treated with a dexamethasone based induction regimen and frontline ASCT. Using the HistoRx PM-2000 quantitative immunohistochemistry platform, coupled with the AQUA analysis software, we have examined the expression of the following proteins: FGFR3 which is associated with t(4;14), cyclin B2 and Ki-67 which are associated with cellular proliferation, TACI which is associated with maf deregulation, and phospho-Y705 STAT3 and p65NF-κB, which are associated with myeloma cell growth and survival. For FGFR3, patients were divided into FGFR3 positive and negative groups based on hierarchical clustering of their AQUA score. For all other proteins examined, based on AQUA scores, the top quartiles or quintiles of patients were classified as high expression groups. Based on the univariate analysis, patients were further classified as “High Risk” MM if they had been identified as high expressers of either TACI, p65NF-κB or FGFR3. The Kaplan-Meier method was used to estimate time to progression and overall survival. Multivariate analysis was performed using the Cox regression method. Results: 82 patients were included in this study. In univariate analysis, FGFR3 and p65NF-κB expression were associated with significantly shorter TTP (p=0.018 and p=0.009) but not OS (p=0.365 and p=0.104). TACI expression levels predicted for worse OS (p=0.039) but not TTP (p=0.384). High expression of Ki67 or phospho-Y705 STAT3 did not affect survival. Of the 82 cases, 67 were included in the multivariate analysis since they had AQUA scores available for all markers: 26 (38.8%) were considered as High Risk by their AQUA scores and had significantly shorter TTP (p=0.014) and OS (p=0.006) compared to the Low Risk group. The median TTP for the Low and High Risk groups was 2.9 years and 1.9 years, respectively. The 5-years estimates for OS were 60.6% for the High Risk group versus 83.5% for the Low Risk group. Multivariate analysis was performed using del13q and our risk group classification as variables. Both our risk group classification and del13q were independent predictors for TTP, having 2.4 and 2.3 greater risk of relapse, respectively. Our risk group classification was the only independent predictor of OS with the High Risk group having a 5.9 fold greater risk of death. Conclusions: We have found that the expression of FGFR3, TACI, and p65NF-κB, in an automated and fully quantitative tissue-based array, is a powerful predictor of survival post-ASCT in MM and eliminates the “observer-dependent” bias of scoring TMAs. A validation of this “High Risk” TMA based signature is currently underway in larger and independent cohorts. Figure Figure
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.002 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.002 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.000 |
| Open science | 0.001 | 0.000 |
| Research integrity | 0.001 | 0.000 |
| Insufficient payload (model declined to judge) | 0.002 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".