MedicalResearch.com: Medical Research Exclusive Interviews March 17 2015
Thesis Presentation-Abbreviated version
1. Modeling Survival After Surgery On
The Metastatic Spine
Ahmer Ghori MD, Dana Leonard BA, Andrew Schoenfeld MD, Ehsan Saadat MD, Mitch B Harris MD
Ahmer K Ghori MD
Harvard University Orthopaedic Surgery Program
2. My GoalToday
Review the past and present state of
managing metastatic spine.
Discuss how our model may improve the
present state.
3. Historical: No advantage with surgery
Present: Surgery helps
Limitation: Can’t predict post op survival
Who to operate on?
Our Model
Future: Patient specific management
Evolution of
spinal
instrumentation
Patient
Specific
Variables
Predict survival
reliably
4. Historic Management:Why no surgery ?
Laminectomy was the only surgical option.
No difference in laminectomy + radiation versus
radiation alone with regard to:
pain relief
ambulation
sphincter function.
Young et al 1980
6. Modern Literature
Surgery helps in managing metastatic spine
Pain.
Ambulation.
Maintain independence at end of life.
Direct decompressive surgical resection in the treatment of spinal cord compression caused by metastatic cancer: a randomised trial. The Lancet.
2005;366(9486):643–648. doi:10.1016/S0140-6736(05)66954-1.
7. Present Limitations
Choosing surgical candidates
Not a reliable process.
Existing models:
Do not predict survival reliably !
Correlation with survival 20-50 %
Ragel, B. T. et al. Life Expectancy and Metastatic Spine Scoring Systems: An Academic Institutional Experience. J Spinal Disord Tech
(2014).
Wang M, Bünger CE, Li H, Wu C, Høy K, Niedermann B, et al. Predictive value of Tokuhashi scoring systems in spinal metastases,
focusing on various primary tumor groups: Evaluation of 448 patients in the Aarhus spinal metastases database. Spine 2012;37: 573-
582.
12. Existing scoring systems do not
capture patient’s overall health state.
Ghori et al 2015
Why can’t we predict survival ?
13. Let’s consider a hypothetical scenario
Bauer Score: 4
Can we predict post op
survival from a score of 4
alone?
Consider 2 patients
Female
• Breast cancer
• No visceral metastases
• Solitary spine metastasis
18. Both patients have a Bauer score of 4 !
Differences will affect survival
Need a model that is more specific.
Existing Models: Shortcoming
VS.
20. Methods
Retrospective review at four participating
institutions (BIDMC, BWH, MGH, Dartmouth).
Identify patients who had undergone spine surgery
for metastatic disease between 2007 and 2013.
Over 300 patients across 5 years included.
Ghori et al 2015
21. Variables we studied
Main predictors of survival we studied
Pre-operative modified Bauer score
Pre- operative ambulatory status
Pre-operative serum albumin level.
Pre-operative serum creatinine level.
Primary tumor type.
Demographics: age, sex, race.
Number of comorbidities.
We coined these “healthiness states”
Ghori et al 2015
22. Ghori et al 2015: Our Model
Analyzed all permutations of these three variables.
Chose combination which has best correlation with survival.
High Bauer score (3,4): 2 points
Intact ambulatory status: 1 point
Normal serum albumin level : 1 point
Ceiling score of 3.
Ghori et al 2015
24. Ghori et al 2015
Our Model was able to explain 74% of the variation
in 1-year survival.
What does that mean?
In contrast existing scoring systems : 20- 50 %
correlation with survival
Ragel, B. T. et al. Life Expectancy and Metastatic Spine Scoring Systems: An Academic Institutional Experience. J Spinal Disord Tech (2014).
Wang M, Bünger CE, Li H, Wu C, Høy K, Niedermann B, et al. Predictive value of Tokuhashi scoring systems in spinal metastases, focusing on various
primary tumor groups: Evaluation of 448 patients in the Aarhus spinal metastases database. Spine 2012;37: 573-582.
26. High modified Bauer
score and
1. Intact ambulation
2. Normal albumin
78% survival
Modified
Bauer Score
(4):
59% survival
Case 1:Bauer 4
27. High modified Bauer
score BUT
1. CANNOT ambulate
2. Low albumin
32% survival
Modified
Bauer Score
(4):
59% survival
Case 2:Bauer 4
28. Ghori et al 2015
Modified Bauer Score (4): 59%
No additional heath states: 30 %
Modified Bauer Score (4): 59%
Intact Ambulation/Normal albumin: 78 %
29. Ghori et al 2015
30% Vs. 78% survival: big difference !
Capture patient health state Tell apart patients predict survival reliably
30. Our Model: Advantages
Simple to apply.
Superior to existing models in predicting survival.
300 patients in our analysis:
one of the largest cohorts in literature.
31. Our Model: Limitations
Retrospective nature of our data.
Model developed from surgical patients so may not
be extrapolated to non-surgical patients.
32. Future Direction
Studies to validate our model by applying it to:
Different databases.
Prospectively studying it.
34. Historical: No benefit with Surgery
Present : Surgery – too much ?
Who to operate on?
Our Model
Future approach: Improve patient selection
Operate on the RIGHT PATIENT
Evolution of
spinal
instrumentation
Include Patient
“Health State”
Predict survival
reliably