G. Lahat, B. Wang, D. Tuvin, DA. Anaya, C. Wei, B. Bekele, KD. Smith, AJ. Lazar, PW. Pisters,

Clinical variables, pathological factors, and molecular markers for enhanced soft tissue sarcoma prognostication

G. Lahat, B. Wang, D. Tuvin, DA. Anaya, C. Wei, B. Bekele, KD. Smith, AJ. Lazar, PW. Pisters,

RE. Pollock, D. Lev

Sarcoma Research Center

UT MD Anderson Cancer Center

Houston, TX U.S.A.

Daniel Tuvin

Good afternoon. I would like to thank the program commitee for the opportunity to present our data at this meeting. there are no disclosues.

Daniel Tuvin

We examined several prognostic criteria that are not currently part of the AJCC STS staging system. Our results show that incorporation of these factors into this system may help to better describe prognosis.

Current STS staging systems have several important shortcomings

• The TNM and grade criteria do not reflect the heterogeneity of STS

• Available nomograms are not universally applicable

• No current STS staging system includes molecular predictors of outcome

• Patient I- 6cm high grade extremity UPS

• Patient II- 25cm retroperitoneal dedifferentiated LPS

Both are AJCC Stage III patients!

Purpose

To identify clinical, pathological, and molecular descriptors of STS clinical behavior for inclusion in revised staging systems

Daniel Tuvin

Our hypothesis was that current AJCC staging criteria need further evaluation and our purpose was to identify and validate them.

Methods

• UTMDACC STS prospective database

• Univariate and multivariate statistical analyses

• Clinically annotated STS tissue microarray

Daniel Tuvin

And for this purpose 1. We reviewed the MDA prospective STS database 2. Univariate and multivariate anlyses were performed. 3. In addition we created a TMA using more than 250 selected specimens.

Local recurrence only 474Primary & Metastatic 435Metastatic disease only 350

UTMDACC STS Database1996-2007

6,702 patients

Definitive treatment3,717

Primary disease only2,458

Surgical treatment of primary tumor

1,442

Study cohort1,091

Second opinion2,985

R2 resection and non-specific histologies

351

Non- surgical treatment

1,016

Patient and tumor characteristics

Male: 52%; Female: 48%

Median age (range): 54.5 years (15-91)

Median follow-up: 53.3 months

Extremity 58%

Intra thoracic

6%

Intra abdominal

26%

Superficial trunk 6%

Head and neck 2.5%

Non-extremity location is associated with increased STS-specific mortality

Non-extremity (n= 464; 42%)

Extremity (n= 627; 58%)

p<0.0001

Daniel Tuvin

This graph shows that non-extremity location, which was mainly retroperitoneal in this cohort is a significant independent risk factor for sarcoma specific mortality.

A T3 category may be added to the AJCC STS staging system

p<0.0001

Tumor size>15cm (n= 230; 21%)

Tumor size 5-10cm (n=336; 31%)

Tumor size 10-15cm (n= 202; 19%)

Tumor size< 5cm (n= 309; 29%)

Daniel Tuvin

we also evaluated whether further stratification of tumor size into more categories was more informative than the T1/ T2 binomial criteria of the AJCC STS staging system.

Daniel Tuvin

Unlike the Royal Marsden group we did not find a difference between 5-10 and 5-15 cm tumors; however our data demonstrate that size larger than 15cm is associated with significant increase in disease related mortality.

High grade is associated with increased STS-specific mortality

p<0.0001

High grade (n=737; 67.6%)

Low/intermediate grade (n=354; 32.2%)

Daniel Tuvin

We stratified grade into 2 categories; low and high grades. As expected grade was found to be a significant risk factor for STS related mortality.

Interaction between variables: tumor size and grade effects on STS-specific

mortality

p<0.0001

High grade, size>15cm

High grade, size 5-15cm

High grade, size<5cm

Low/intermediate grade

Daniel Tuvin

Interestingly we found that size does matter only for high grade STS; whereas it does not affect the survival of patients with low grade tumors.

Low /intermediate grade, negative margins

Low/intermediate grade, positive margins

High grade, negative margins

High grade, positive margins

Interaction between variables: margin positivity and grade effects on STS-

specific mortality

p<0.0001

Daniel Tuvin

As for tumor size matgins positivity is a critical prognostic factor only for HG STS.

Multivariate Cox Proportional Hazard Models for STS-specific mortality

Variable Levels HR P value

Microscopic margins Positive vs. negative 5.9 <0.0001

Primary site Non-extremity vs. extremity 3.19 <0.0001

Tumor size 5-15cm vs. <5cm >15cm vs. <5cm

2.987.45

<0.0001<0.0001

Disease grade High grade vs. low/intermediate grade 2.06 <0.0001

Histology UPS vs. WDDedifferentiated liposarcoma vs. WDOthers vs. WD

8.074.022.98

0.0010.005<0.0001

Multivariate Cox Proportional Hazard Models for STS local recurrence free

survival


Age Continuous in 10 years increment 1.26 <0.0001

Primary site Non-extremity vs. extremity 1.84 <0.0001

Margin positivity Positive vs. negative margins 2.39 <0.0001

Disease grade High grade vs. low/intermediate grade 1.90 0.001

Gender Male vs. female 1.54 0.01

ConclusionsAnn Surg Oncol 2008; 15:2739-48

STS size, site, grade, histology, and microscopic margin status should be included in a revised staging system

Can we further improve and individualize prognostication?

IDAN LAHAT

We conclude that tumor size and site, grade, histology and microscopic margins should be included in a revised STS staging system. Molecular markers will probably augment every staging system in the future. Recurrence during post operative follow-up which is a dynamic factor is the most significant prognstic factor for adverse outcome.We are in the process of creating a dynamic staging system which includes dynamic factors such as time from initial diagnosis and recurrence.

Every STS is “unique”

CureDistant

metastasis followed by

death

6cm, extremity, HG, UPS,

R0 resection

6cm, extremity, HG, UPS,

R0 resection

Patient A Patient B

Clinical and pathological prognostic factors are not enough!

Molecular markers are important potential prognostic factors

• High throughput assays

• Detection of DNA, RNA, and protein targets

• Simultaneous analysis of large tumor sets

• Correlation with clinical data

TMA (n=205)TMA (n=205)Growth and metastasis

Ki-67

Apoptosis/survival

P53

MDM2

Bcl2

Bcl-x

Cytokines/receptors/signaling

EGFR

VEGF

β-Catenin

Extracellular matrix

MMP2

MMP9

guy lahat

In our era high throuput assays such as TMA can be used for detection of DNA, RNA and protein targets in a large set of specimens examined at the same time. We created a TMA of 257 HG STS. So far we stained them for this list of MOLECULAR targets. All have been shown to be associated with sarcoma progression in small series.

High MMP2 expression correlates with decreased STS-specific survival

Overall Survival time (month)

Pro

po

rtio

n S

urv

ivin

g

0 12 24 36 48 60 72 84 96 108 120 132 144 156

0.0

0.2

0.4

0.6

0.8

1.0

P-value= 0.004

Percent of area mmp2 positive <=10% ( E / N = 23/101 )Percent of area mmp2 positive >10% ( E/N =35/67)

72%

46%

Disease specific survival time (months)

Percent MMP2 pos ≤ 10%

Percent MMP2 pos > 10%

Daniel Tuvin

of them, mmp2 (matrix metaloproteinase 2) which is a positive regulator of cell migration was a significant prognostic factor for a decreased survival.

Multivariate Cox Proportional Hazard Models for disease specific survival

(TMA cohort)


Primary site Non-extremity vs. extremity 2.95 0.001

Disease grade High grade vs. low/intermediate grade 2.5 0.02

Age Continuous in 10 years increment 1.62 <0.0001

MMP2 expression > 10% vs. ≤ 10% 1.74 0.04

Multivariate Cox Proportional Hazard Models for local recurrence

(TMA cohort)


Age Continuous in 10 years increment

1.51 0.008

Primary site Non-extremity vs. extremity 4.09 0.005

MMP2 expression > 10% vs. ≤ 10% 6.28 0.006

Matrix metalloproteinases(MMP2) and STS

Number of patients Correlation with outcome

Maguire PD, et al (Oncology, 2000) 12 Negative

Benassi MS, et al (Ann Oncol, 2001) 73 Positive; grade, DFS

Previous series

Conclusions

• High MMP2 expression may be an adverse independent predictor of outcome in STS

• Inclusion as a molecular prognostic factor in future STS staging systems, pending large scale validation

• Individualized therapeutic strategy

• Should be further studied as potential targets for therapy

Acknowledgments

Vision and direction: Insights and teamwork:

Sarcoma Research CenterUniversity of Texas

MD Anderson Cancer Center

Dina Lev, MD Daniel Tuvin, MD

Raphael Pollock, MD, PhD Daniel Anaya, MD

Peter Pisters, MD Kerrington Smith, MD

Alex Lazar, MD, PhD

Nebiyou Bekele, PhD

Kevin Coombs, PhD

Caimiao Wei, PhD

Daniel Tuvin

A multidiciplinary group of people were gathered to work on this project. I want to thank them all, and especially to Daniel Tuvin who spent long hours creating the database Dr Pollock and Dr Lev who for mentoring me since my arrival to MDA.

Thank you for your attention!

TMA Tumor characteristics

Extremity 86%

Size> 5cm (76%)

HG (75%)

UPS

78%

Negative margins (70%)

G. Lahat, B. Wang, D. Tuvin, DA. Anaya, C. Wei, B. Bekele, KD. Smith, AJ. Lazar, PW. Pisters,

Documents