1. Prognostication in MF: From CBC to cytogenetics to molecular markers
Alessandro M. Vannucchi
University of Florence, Italy

2. Survival is Significantly Shortened in PMF
Median survival: 4.6 versus 6.5 y
Cervantes F et al. JCO 2012; 24:

3. Why do we Need Accurate Prognostic Scores?
Long term remissions with the potential of being cured have been described only in patients undergoing allogeneic HSCT
However, HSCT approach to MF has several limitations, currently:
Disease related
Procedure related
Median age is 65 y (but changing…)
Donor availability
Age-related comorbidities
Only one prospective trial (+1 not yet fully reported)
Fragile patients (splenomegaly, cytopenias, previous therapies…)
High, and highly variable, TRM
Rapid evolution to AML
Choice of fully ablative versus RIC
Uncertainties about whom and when
Poor information about long-term effects

4. Risk Stratification in PMF
Variable
IPSS
DIPSS
Age >65 y 
Constitutional symptoms
Hemoglobin <10 g/dL
Leukocyte count >25x109/L
Circulating blasts > 1%
Platelet count <100x109/L
RBC transfusion need
Unfavorable karyotype
+8,-7/7q-,i(17q),inv(3), -5/5q-,12p-, 11q23 rearr.
Cervantes et al, Blood 2009;113:
Passamonti et al, Blood 2010; 115:1703-8
Gangat N et al, J Clin Oncol 2011; 29:392-7

5. International Prognostic Scoring System-IPSS
Points
 Median survival (mo)
Low
135
Int-1 1 95
Int-2 2 48
High
>3 27
Low
Int-1
Int-2
High
Cervantes F et al. Blood 2009;113:

6. Impact of Anemia on Disease Progression
Disease-related anemia* has been associated with worse prognosis in all
risk scoring systems (Lille, IPSS and derivatives)
RBC transfusional dependency is included in the DIPSS-plus score
* NOT treatment-related
Passamonti F.et al, Blood 2010; 115:1703-8

7. Dynamic IPSS (DIPSS) Points Median survival (mo) Low Not reach. Int-1
Not reach.
Int-1
1-2
170
Int-2
3-4 48
High
5-6 18
Passamonti F et al. Blood 2010;115:1703-8

8. Survival by Cytogenetic Category in PMF
At diagnosis
Beyond initial diagnosis
Survival by cytogenetic category. (A) Patients evaluated at diagnosis. (B) Patients evaluated beyond initial diagnosis. CPX indicates complex (≥ 3) abnormalities.
CPX= complex (>3 abnormalities)
Tam C S et al. Blood 2009;113:

9. "Unfavorable" Karyotype in PMF: Effect on OS
- Complex karyo,
- Sole or 2 abnormalities including:
Trisomy 8
-7/del(7q)
Del(5q)
Inv(3)
isochromosome 17q/17p-
12p-
11q23 abnormality
Unfavorable Karyo
M 2.0 yr
5-yr survival: 8%
Favorable Karyo
M 5.2 yr
5-yr survival: 51%
Caramazza D et al. Leukemia 2011; 25:82-88

10. "Unfavorable" Karyotype in PMF: Effect on LFS
5-yr AML transformation rate: 7%
Unfavorable Karyo
5-yr AML transformation rate: 46%
Caramazza D et al. Leukemia 2011; 25:82-88

11. Risk Stratification in PMF
Variable
IPSS
DIPSS
DIPSS-plus
Age >65 y 
Constitutional symptoms
Hemoglobin <10 g/dL
Leukocyte count >25x109/L
Circulating blasts > 1%
Platelet count <100x109/L
RBC transfusion need
Unfavorable karyotype
+8,-7/7q-,i(17q),inv(3), -5/5q-,12p-, 11q23 rearr.
Cervantes et al, Blood 2009;113:
Passamonti et al, Blood 2010; 115:1703-8
Gangat N et al, J Clin Oncol 2011; 29:392-7

12. Dynamic IPSS-plus (DIPSS-plus)
Survival
Months
∆ Low risk (0 adverse points)
n=66; median survival ~ 185 months
▲ Intermediate-1 risk (1 adverse point)
n=174; median survival ~ 78 months
○ Intermediate-2 risk (2 or 3 adverse points)
n=360; median survival ~ 35 months
High risk (4 or more adverse points)
n=193; median survival ~ 16 months
Low risk (0 variable)
M 185 mo
Int-1 risk (1 variable)
M 78 mo
Int-2 risk (2-3 variables)
M 35 mo
High risk
(>4 variables)
M 16 mo
Gangat et al. J Clin Oncol 2011;29:392-7

13. Adverse Impact of Monosomal Karyotype
Normal karyo
M 50 mo
Monosomal
karyo
M 6 mo
Sole +8
M 20 mo
Complex karyo
No monosomal
M 24 mo
Vaidya R et al. Blood 2011;117:

14. “Very-High Risk” Patients: >80% Mortality Very-High risk variables
At 2 Years
Very-High risk variables
monosomal karyotype
inv(3)/i(17q)
or any 2 of the following:
PB blasts >9%
WBC >40x109/L
other unfavorable karyotype
Low (3%)
Int-1 (11%)
Int-2 (26%)
High (53%)
Very High (82%)
N=884 pts at Mayo. 564 deaths, corresponding to 64%; 60 cases of AML l8% total population)
The two strongest individual variables were MoKary and inv(3)/I(17q). The other predicted >80% mortality when present in at least two.
The Very-High risk category included 52 pts vs 298 of high-risk. Of these 42 have been assigned to the HR and 10 to the Int-2 risk accoridng to DIPPSplus
Leukemia risk was 31% vs 7%.
Tefferi A et al. Blood 2011; 118:4595-8

15. Novel Prognostic Variables
Somatic Mutations
Germline Characteristics
Cytokines
Other biomarkers (FLC, hepcidin & ferritin levels)

16. JAK2 V617F Mutation and Prognosis in PMFWT
V617F
JAK2 V617F
JAK2 wt
89 (31.1%)
68 (34.5%)
103.4 mo
( )
137.6 mo
( )
N=483
Guglielmelli P et al, ASH2012

17. No Impact of JAK2V617F on Leukemia RiskWT
V617F
P=0.839
HR: 1.05 (95% CI, )
(Competitive risk analysis)
Guglielmelli P et al, ASH2012

18. Blast transformation: Kaplan-Meier analysis
IPSS prognostic score
4.33 ( )
0.007
JAK2 V617F wt or Homo vs. Hetero
2.43 (1.44-4)
0.02
Barosi et al. PlosOne 2013, In press

19. Mutation Complexity in PMF
382 (79.1%) of patients presented at least one somatic mutation
154 pts (32.5%) had >2 mutations
31 pts (6.4%) had >3 mutations
Guglielmelli P et al, ASH2012

20. Mutations Associated with Reduced Overall Survival
in Multivariate Analysis
EZH2
ASXL1 WT WT
Mut
Mut P=
P<
SRSF2
Hazard Risk
(95% CI range) P
EZH2
1.91 ( )
0.025
ASXL1
2.21 ( )
<0.0001
SRSF2
2.60 ( ) WT
Mut
P<
Guglielmelli P et al, ASH2012

21. Mutations Associated with Leukemia in Multivariate Analysis
EZH2
ASXL1 WT
Mut
Mut WT
P=.003
HR=1.98 (95%CI: )
P<.0001
HR=2.5 (95%CI: )
SRSF2
IDH1/2
Mut
Mut WT WT
P=.007
HR= 2.73 (95%CI: )
P<.0001
HR= 2.66(95%CI: )
* Competitive Risk Analysis
Guglielmelli P et al, ASH2012

22. A "Molecularly High-Risk" Status Associates with Reduced Overall Survival
Low Risk
High Risk
P<0.0001
EZH2
ASXL1
SRSF2
IDH1/2
HR= 2.29 (95%CI: )
In the “molecularly high-risk” category, overall survival was 81 months (range: ) compared with 148 months (range: ) in the “molecularly low-risk” category (P<0.0001). Mutivariate analysis.
Guglielmelli P et al, ASH2012

23. A "Molecularly High-Risk" Status Associates With Leukemia Transformation
EZH2
ASXL1
SRSF2
IDH1/2
Low Risk
P<0.0001
HR (95%CI: )
In the “molecularly high-risk” category, leukemia-free survival was 129 months (range: ) compared with 323 months (range: ) in the “molecularly low-risk” category (P<0.0001) – Competitive risk analysis
Guglielmelli P et al, ASH2012

24. The "Molecularly High-Risk" Status Contributes to Refined IPSS Categorization
IPSS (LOW-INT1)
IPSS (INT2-HIGH)
P= 0.017
P= 0.002
Low Risk
Low Risk
High Risk
High Risk
Molecular Risk category
Median Overal Survival, months (range)
LOW
264.2
( )
HIGH
125.6
( )
Molecular Risk category
Median Overal Survival, months (range)
LOW
71.5
( )
HIGH
32.4
( )
Guglielmelli P et al, ASH2012

25. Hazard Risk (95%CI,range)
The "Molecularly High-Risk" Status Predicts for Leukemia Risk within IPSS Categories
IPSS (LOW-INT1)
IPSS (INT2-HIGH)
P= 0.001
P= 0.01
High Risk’
High Risk
Low Risk
Low Risk
IPSS category
Hazard Risk (95%CI,range)
LOW-INT1
2.28 ( )
INT2-HIGH
3.22 ( )
Guglielmelli P et al, ASH2012

26. ABSTRACT #430
Prognostic Interactions Between SRSF2, ASXL1, and IDH Mutations in Primary Myelofibrosis and Determination of Added Value to Cytogenetic Risk Stratification and DIPSS-Plus
Terra L Lasho, MT, (ASCP)1*, Naseema Gangat, MD1*, Christy Finke, BS1*, Rebecca R. Laborde, PhD1*, Curtis A Hanson, MD2*, Rhett P Ketterling, MD3*, Ryan A Knudson3*, Animesh Pardanani, MBBS, PhD1 and Ayalew Tefferi, MD1

27. The A3669G Polymorhism of Glucocorticoid Receptor Contributes to Blast Transformation in PMF
The A3669G allele is a susceptibility allele for PMF (HR )
The G/G allele associated with a «more-myeloproliferative» phenotype
OS*
BT*
0.47 per 100 pt-yr
N=274
The A3669G allele in the untranslated region of human GR stabizes the mRNA of the dominant-negative Beta isoform of the GR, and contributes to development of erythrocytosis in PV where it is represented at highre frequency than in normal subjects. The variant has been associated with an excess proliferative signaling in hematopoietic cells.
This study involved 499 PMF
GG associated with higher WBC, larger spleen, higher CD34+ in PB (only the latter was JAK2V617F independent in multivariate)
The association with BT remained significant after correction with know risk factors for AML (age, sex, WBC, IPSS, JAK2 homozygosity :HR 3.3, P=0.006)
N=21
77.6mo vs 298mo; P=0.049
13.6 per 100 pt-yr
76.7mo vs 261mo; P=0.018
remained significant in multivariate
*, restricted to JAK2V617Fpos pts, n=295
Barosi G et al, Blood 2012; 120:3112-7

28. Abnormally Increased IL-8 and IL2R Plasma Levels Are Prognostically Detrimental
Int-1 (n=27)
All (n=127)
Survival data of (A) all 127 patients and (B) 90 treatment-naive patients with primary myelofibrosis stratified by the presence or absence of interleukin 8 (IL-8) and IL-2R levels that exceed three standard deviations above the normal mean.
Survival data of (A) 27 intermediate-1–risk patients and (B) 70 intermediate-2–risk patients with primary myelofibrosis stratified by the presence or absence of interleukin 8 (IL-8) and IL-2R levels that exceed three standard deviations above the normal mean. Risk categorization is according to the Dynamic International Prognostic Scoring System plus model.5
Int-2 (n=70)
Treatment naive
(n=90)
Tefferi A et al. JCO 2011;29:

29. Plasma Free Light Chain (FLC) Levels Predict Survival in PMF
FCL
FCL +/- IL-8 and/or IL-2R
< 3.78 mg/dL
Both normal
Both abnormal
> 3.78 mg/dL
Either abnormal
Both abnormal
Overall survival Kaplan-Meier survival curves of patients with primary myelofibrosis stratified by plasma free light chain (FLC) concentration above or below the receiver operating characteristic analysis–derived cutoff (3.78 mg/dL). (A) Patients with primary myelofibrosis with FLC more than 3.78 mg/dL had significantly inferior survival compared with those with an FLC level ≤ 3.78 mg/dL (log-rank P < .001). This association with inferior overall survival was sustained on multivariate analysis that considered other standard prognostic variables. (B) Patients with Dynamic International Prognostic Scoring System (DIPSS) -plus high and intermediate-2 risk stratified by DIPSS-plus and the FLC threshold of 3.78 mg/dL. (C) Composite risk model based on total FLC more than 3.78 mg/dL and interleukin-2 receptor (IL-2R) and/or IL-8 levels more than three standard deviations: both normal, either abnormal, and both abnormal. This refined risk model was predictive for survival independent of DIPSS-plus: either abnormal (hazard ratio, 2.1; P = .04) and both abnormal (hazard ratio, 3.7; P <= .001).
Levels above or below the ROC cutoff (3.78 mg/dL)
Pardanani A et al. JCO 2012;30:

30. Conclusions
Prognostic scores in OMF are needed for therapeutic choices
At present, they are mainly reserved for HSCT decision
Most used scores are based on clinical and hematologic variables
Cytogenetics has been show to add to clinical scores
Molecular characterization may help refine clinical scores, be cost-effective and overcome technical limitations of conventional cytogenetics

31. Acknowledgments Section of Hematology, University of Florence
Paola Guglielmelli
Flavia Biamonte
Tiziana Fanelli
Ambra Spolverini
Maria Chiara Susini
Giada Rotunno
Alessandro Pancrazzi
Lisa Pieri
Contributors
Mario Cazzola - Pavia
Gianni Barosi - Pavia
Francisco Cervantes - Barcelona
Andrea Reiter - Mannheim
Andrew Duncombe - Southampton
Katerine Zoe - Athens
Nick Cross - Salisbury