1. Hierarchical Cluster analysis of Florescent in Situ Hybridisation in newly diagnosed myeloma patients Ieuan Walker BSc. (hons) MSc. 4.11.15

2. Introduction Cytogenetics/FISH is a key part of myeloma risk stratification and has recently been included in R-ISS Clear associations between individual FISH abnormalities and effect on outcome e.g. 17p deletion Co-associations between individual abnormalities recognised but less well defined Single centre, retrospective review of FISH abnormality clustering and outcome

3. Methods Patients and Sample analysis Retrospective analysis of all newly diagnosed symptomatic myeloma patients between 2009 and 2014 153 patients with median follow-up 36 months Heterogeneous 1 st line treatment Standard FISH panel (see next slide) Hierarchical cluster analysis for abnormality grouping Effects on overall outcome and relationship with ISS and R-ISS staging systems Hierarchical cluster analysis Using SPSS we formed a similarity score by making each of the cytogenetic lesions proportional to the ‘square Euclidian distance’ We then merged these in turn based on how ‘similar’ they using an algorithm we designed. Survival curves were generated using a standard Kaplan Meier method, and P values by a log-rank test

4. Cytogenetic AbnormalityGenes of interest at Loci Ch1q21 CKS1B, PMSD4 Ch1p32.3CDKN2C Ch5p15.2CSF1R Ch11q13CCND1 Ch11q22.3ATM Ch13q34LAMP1 ch14q23ReIgH Heavy chain t[4:14]IgH/FGFR3 t[6:14]IgH/CCND3 t[11:14}IgH/CCND1 t[14:16]IgH/MAF t[14;20]IgH/MAFB ch17p13.TP53 CEP9triHyperdiploid Chromosome CEP12TriHyperdiploid Chromosome CEP13Monosomy or Hyperdiploid CEP15Hyperdiploid Chromosome 1q21 Gain of Copy Rearranged 11:14 FISH panel

5. Hierarchical cluster analysis. 1 1 Ch1q21 Tri Ch13 monCLUSTER 1 IgH Re Ch1q21 Tri Ch13 monCLUSTER 1 IgH Re Ch5p15.2 Tri CEP9 Tri CEP15 TriCLUSTER 2 Ch11q22.3 Tri Ch5p15.2 Tri CEP9 Tri CEP15 TriCLUSTER 2 Ch11q22.3 Tri Dendrogram using Linkage by Patient

6. What do the clusters mean in terms of prognosis? Median overall survival for our total population has not been reached. Median overall survival for Cluster 1 patients = 42 months Median overall survival for Cluster 2 Not reached Median OS for ISS 1: Not Reached 2: 45 Months 3: 40 months

7. Can we use clusters to provide any additional stratification to the ISS staging criteria? Patients categorized according to their cluster status (1 or 2) and ISS stage 3 Survival benefit for cluster 2 Patients with lesions in both clusters were excluded from the analysis Cluster 1 + ISS3 (n=19) Cluster 2 + ISS 3 (n= 23) ISS 3 (n= 65) Median OS Cluster 1 + ISS 3 = 27m Cluster 2 + ISS 3 Not reached Stage 3 = 40m * * p=0.047 Cluster 1 Ch1q21 Tri Ch13 mon IgH Re Cluster 2 Ch5p15.2 Tri CEP9 Tri CEP15 Tri Ch11q22.3 Tri

8. Does cluster analysis complement the revised ISS? 109 patients restaged according to R-ISS (n =28, 60, 19 respectively) ISS stage 3 patients also analysed according to cluster 1 or 2 (black solid and purple) ISS-3+Cluster 1 has worse prognosis than R-ISS3 Median OS survival: ISS3 + Cluster 1 = 20 months R-ISS3= 23 Months ** ** P<0.005

9. Conclusions FISH can be used as a marker of prognosis but needs to be used in conjunction with biomarkers to identify the poorest prognostic groups. Specific cytogenetic lesions undoubtedly are important prognosticators for survival such asTP53 loss. The FISH “signatures” is also important. While poor prognosis markers counteract positive lesions, the absence of any cluster 2 abnormalities appears to convey even higher risk. Using our unbiased cluster analysis we show that cytogenetics lesions group together. To further validate our algorithm we need to expand the data set… watch this space!

10. Thanks to Guy’s And St Thomas Hospital Haematology Dr Matt Streetly Dr George Double Dr Majid Kazmi Dr Ines El-Najir Grace Miller King’s College Hospital Haematology Prof Steve Schey Viapath (cytogenetics at Guys Hopsital) Dr Michael Neat