HARVARD MEDICAL SCHOOL

29
Chemoradiation in T2 Chemoradiation in T2 Rectal Cancer Based on Rectal Cancer Based on Mesorectal Lymph Node Mesorectal Lymph Node Size: A Decision Analysis Size: A Decision Analysis Informed by Patient Informed by Patient Outcomes Outcomes Chang, Connie Y., M.D., Pandharipande, Pari, Chang, Connie Y., M.D., Pandharipande, Pari, M.D., M.P.H., Harisinghani, Mukesh, M.D., M.D., M.P.H., Harisinghani, Mukesh, M.D., Gazelle, G. Scott, M.D., M.P.H., Ph.D. Gazelle, G. Scott, M.D., M.P.H., Ph.D. HARVARD MEDICAL SCHOOL

description

Optimizing Triage to Preoperative Chemoradiation in T2 Rectal Cancer Based on Mesorectal Lymph Node Size: A Decision Analysis Informed by Patient Outcomes. Chang, Connie Y., M.D., Pandharipande, Pari, M.D., M.P.H., Harisinghani, Mukesh, M.D., Gazelle, G. Scott, M.D., M.P.H., Ph.D. HARVARD - PowerPoint PPT Presentation

Transcript of HARVARD MEDICAL SCHOOL

Page 1: HARVARD  MEDICAL SCHOOL

Optimizing Triage to Optimizing Triage to Preoperative Preoperative

Chemoradiation in T2 Chemoradiation in T2 Rectal Cancer Based on Rectal Cancer Based on Mesorectal Lymph Node Mesorectal Lymph Node Size: A Decision Analysis Size: A Decision Analysis

Informed by Patient Informed by Patient OutcomesOutcomes

Chang, Connie Y., M.D., Pandharipande, Pari, M.D., Chang, Connie Y., M.D., Pandharipande, Pari, M.D., M.P.H., Harisinghani, Mukesh, M.D., Gazelle, G. M.P.H., Harisinghani, Mukesh, M.D., Gazelle, G.

Scott, M.D., M.P.H., Ph.D.Scott, M.D., M.P.H., Ph.D.

HARVARD MEDICAL SCHOOL

Page 2: HARVARD  MEDICAL SCHOOL

Background: Rectal Background: Rectal CancerCancer

MRI has had increasing MRI has had increasing role in preoperative role in preoperative planning for rectal cancer planning for rectal cancer (T-stage)(T-stage)

Large degree of overlap of Large degree of overlap of size of normal/reactive size of normal/reactive and cancer-containing and cancer-containing peri-rectal lymph nodes.peri-rectal lymph nodes.

Page 3: HARVARD  MEDICAL SCHOOL

Lymph Node Staging in Rectal Lymph Node Staging in Rectal CancerCancer

Stage T2 rectal cancer

Perirectal lymph nodes

Page 4: HARVARD  MEDICAL SCHOOL

PurposePurpose

To optimize key patient To optimize key patient outcomes in T2 rectal outcomes in T2 rectal cancer by identifying cancer by identifying mesorectal lymph node mesorectal lymph node size criteria for triage to size criteria for triage to preoperative preoperative chemoradiation.chemoradiation.

Page 5: HARVARD  MEDICAL SCHOOL

MethodsMethods

Decision-Analytic Decision-Analytic ModelModel

Model inputs derived Model inputs derived from literaturefrom literature

T2 rectal cancerT2 rectal cancer

Page 6: HARVARD  MEDICAL SCHOOL

MethodsMethods

Stage T2 Rectal Cancer

Treat All Patients with Pre-operative Chemoradiation

Treat If any Mesorectal Lymph Nodes are > 3 mm

Treat If any Mesorectal Lymph Nodes are > 5 mm

Treat If any Mesorectal Lymph Nodes are > 7 mm

No Preoperative Chemoradiation for Any Patients

Page 7: HARVARD  MEDICAL SCHOOL

Methods: Four Disease Methods: Four Disease ScenariosScenarios

True True PositivePositive

(TP)(TP)

False False NegativeNegative

(FN)(FN)

False False PositivePositive

(FP)(FP)

True True NegativeNegative

(TN)(TN)

Page 8: HARVARD  MEDICAL SCHOOL

Methods: Four Disease Methods: Four Disease ScenariosScenarios

Preoperative Preoperative chemoradiatchemoradiat

ionionTrue True

PositivePositive

(TP)(TP)

False False NegativeNegative

(FN)(FN)

No No preoperative preoperative chemoradiatchemoradiat

ionion

False False PositivePositive

(FP)(FP)

True True NegativeNegative

(TN)(TN)

Page 9: HARVARD  MEDICAL SCHOOL

Methods: Four Disease Methods: Four Disease ScenariosScenariosLymph node Lymph node metastases metastases at pathologyat pathology

No lymph No lymph node node metastases at metastases at pathologypathology

Preoperative Preoperative chemoradiatchemoradiat

ionionTrue True

PositivePositive

(TP)(TP)

False False NegativeNegative

(FN)(FN)

No No preoperative preoperative chemoradiatchemoradiat

ionion

False False PositivePositive

(FP)(FP)

True True NegativeNegative

(TN)(TN)

Page 10: HARVARD  MEDICAL SCHOOL

Methods: Four Disease Methods: Four Disease ScenariosScenariosLymph node Lymph node metastases metastases at pathologyat pathology

No lymph No lymph node node metastases at metastases at pathologypathology

Preoperative Preoperative chemoradiatchemoradiat

ionion

Appropriate Appropriate treatment – treatment –

expected expected morbidity of morbidity of preoperative preoperative

chemoradiatiochemoradiationn

No No preoperative preoperative chemoradiatchemoradiat

ionion

Appropriate Appropriate treatmenttreatment

TP FP

FN TN

Page 11: HARVARD  MEDICAL SCHOOL

Methods: Four Disease Methods: Four Disease ScenariosScenariosLymph node Lymph node metastases metastases at pathologyat pathology

No lymph No lymph node node metastases at metastases at pathologypathology

Preoperative Preoperative chemoradiatchemoradiat

ionion

Appropriate Appropriate treatment – treatment –

expected expected morbidity of morbidity of preoperative preoperative

chemoradiatiochemoradiationn

Unnecessary Unnecessary chemoradiationchemoradiation

No No preoperative preoperative chemoradiatchemoradiat

ionion

Appropriate Appropriate treatmenttreatment

TP FP

FN TN

Page 12: HARVARD  MEDICAL SCHOOL

Methods: Four Disease Methods: Four Disease ScenariosScenariosLymph node Lymph node metastases metastases at pathologyat pathology

No lymph No lymph node node metastases at metastases at pathologypathology

Preoperative Preoperative chemoradiatchemoradiat

ionion

Appropriate Appropriate treatment – treatment –

expected expected morbidity of morbidity of preoperative preoperative

chemoradiatiochemoradiationn

Unnecessary Unnecessary chemoradiationchemoradiation

No No preoperative preoperative chemoradiatchemoradiat

ionion

Increased Increased

morbidity of morbidity of

post-operative post-operative chemoradiation, chemoradiation,

increased increased likelihood of likelihood of

local recurrencelocal recurrence

Appropriate Appropriate treatmenttreatment

TP FP

FN TN

Page 13: HARVARD  MEDICAL SCHOOL

Base Case AnalysisBase Case Analysis

SensitiviSensitivityty

SpecificitSpecificityy

All nodes considered All nodes considered malignantmalignant

11 00

Nodes considered Nodes considered malignant if any node malignant if any node > 3 mm> 3 mm

0.910.91 0.430.43

Nodes considered Nodes considered malignant if any node malignant if any node > 5 mm> 5 mm

0.730.73 0.750.75

Nodes considered Nodes considered malignant if any node malignant if any node > 7 mm> 7 mm

0.550.55 0.910.91

No nodes considered No nodes considered malignantmalignant

00 11* From Kim, et al (2004)

Page 14: HARVARD  MEDICAL SCHOOL

Base Case AnalysisBase Case AnalysisPre-Pre-operative operative ChemoradiatChemoradiationion

Post-Post-operative operative ChemoradiaChemoradiationtion

Acute Acute ChemoradiatiChemoradiation Toxic on Toxic EffectsEffects

27%27% 40%40%

Long-term Long-term ChemoradiatiChemoradiation Toxic on Toxic EffectsEffects

14%14% 24%24%

5-year 5-year Probability of Probability of Local Local RecurrenceRecurrence

6%6% 13%13%* Sauer, et al (2004)

Page 15: HARVARD  MEDICAL SCHOOL

Secondary AnalysisSecondary Analysis Individual node radiology-pathology Individual node radiology-pathology

correlation correlation Schnall et al (1994), Brown et al Schnall et al (1994), Brown et al

(2003)(2003) Expanded data (318 nodes from 78 Expanded data (318 nodes from 78

patients)patients) Subject to “clustering bias”Subject to “clustering bias”

USPIO lympUSPIO lymph node contrast agenth node contrast agent Lahaye et al (2008)Lahaye et al (2008)

Page 16: HARVARD  MEDICAL SCHOOL

Sensitivity AnalysisSensitivity Analysis

Performed to assess the impact of Performed to assess the impact of uncertainty in key model uncertainty in key model parameter estimates upon clinical parameter estimates upon clinical outcomesoutcomes

Calculated 95% confidence Calculated 95% confidence intervals for sensitivity and intervals for sensitivity and specificity of each strategyspecificity of each strategy Repeated analysis with upper and Repeated analysis with upper and

lower limits of the confidence lower limits of the confidence intervals.intervals.

Page 17: HARVARD  MEDICAL SCHOOL

Results – Base Case Results – Base Case AnalysisAnalysis

0.00%

5.00%

10.00%

15.00%

20.00%

25.00%

30.00%

Treat All Treat > 3mm Treat > 5mm Treat > 7 mm Treat None

% Patients with Acute Chemoradiation Toxicity% Patients with Long-Term Chemoradiation Toxicity5-Year Local Recurrence

Page 18: HARVARD  MEDICAL SCHOOL

Results – Base Case Results – Base Case AnalysisAnalysis

0.00%

5.00%

10.00%

15.00%

20.00%

25.00%

30.00%

Treat All Treat > 3mm Treat > 5mm Treat > 7 mm Treat None

% Patients with Acute Chemoradiation Toxicity% Patients with Long-Term Chemoradiation Toxicity5-Year Local Recurrence

● Lowest Value

Page 19: HARVARD  MEDICAL SCHOOL

Results – Base Case Results – Base Case AnalysisAnalysis

*

*

*** *

*

0.00%

5.00%

10.00%

15.00%

20.00%

25.00%

30.00%

Treat All Treat > 3mm Treat > 5mm Treat > 7 mm Treat None

% Patients with Acute Chemoradiation Toxicity% Patients with Long-Term Chemoradiation Toxicity5-Year Local Recurrence

● Lowest Value

Page 20: HARVARD  MEDICAL SCHOOL

Results – Base Case Results – Base Case AnalysisAnalysis

0.00%

5.00%

10.00%

15.00%

20.00%

25.00%

30.00%

Treat All Treat > 3mm Treat > 5mm Treat > 7 mm Treat None

% Patients with Acute Chemoradiation Toxicity% Patients with Long-Term Chemoradiation Toxicity5-Year Local Recurrence

● Lowest Value

Page 21: HARVARD  MEDICAL SCHOOL

Results –Results –Secondary and Sensitivity Secondary and Sensitivity

AnalysisAnalysis Individual node analysis – similar pattern Individual node analysis – similar pattern

of results to base case analysisof results to base case analysis Upper limits of all confidence intervals – Upper limits of all confidence intervals –

differed for long-term chemoradiation differed for long-term chemoradiation toxicitytoxicity Minimized if treat no patients preoperativelyMinimized if treat no patients preoperatively

Lower limits of all confidence intervals – Lower limits of all confidence intervals – differed only for acute chemoradiation differed only for acute chemoradiation toxicitytoxicity Minimized if treat patients with LNs > 7 mmMinimized if treat patients with LNs > 7 mm

Page 22: HARVARD  MEDICAL SCHOOL

Results – Sensitivity Results – Sensitivity AnalysisAnalysis

USPIO-EnhancementUSPIO-Enhancement

*

* *0.00%

5.00%

10.00%

15.00%

20.00%

25.00%

30.00%

Treat All Treat >3mm

Treat >5mm

Treat > 7mm

TreatNone

USPIOPositivity

% Patients with Acute Chemoradiation Toxicity% Patients with Long-Term Chemoradiation Toxicity5-Year Local Recurrence

Page 23: HARVARD  MEDICAL SCHOOL

LimitationsLimitations

Reduction of a complex Reduction of a complex disease into a simple disease into a simple decision model.decision model.

Correct identification of Correct identification of stage T2 rectal cancerstage T2 rectal cancer

Page 24: HARVARD  MEDICAL SCHOOL

ConclusionsConclusions

Lymph node size criteria used is Lymph node size criteria used is based on outcome prioritized at based on outcome prioritized at the individual patient levelthe individual patient level Acute toxicity – treat no patientsAcute toxicity – treat no patients Long-term toxicity – treat > 7 mmLong-term toxicity – treat > 7 mm Local recurrence – treat all Local recurrence – treat all

patientspatients A higher threshold may better A higher threshold may better

balance all three outcomes.balance all three outcomes.

Page 25: HARVARD  MEDICAL SCHOOL

ConclusionsConclusions

USPIO-positivity should be USPIO-positivity should be better than all size criteria for better than all size criteria for triaging patients to pre-triaging patients to pre-operative chemoradiation.operative chemoradiation.

Page 26: HARVARD  MEDICAL SCHOOL

Thank you!Thank you!

Page 27: HARVARD  MEDICAL SCHOOL

ReferencesReferences

Brown G, Richards, CJ, Bourne, MW, et Brown G, Richards, CJ, Bourne, MW, et al. Morphologic predictors of lymph al. Morphologic predictors of lymph node status in rectal cancer with use of node status in rectal cancer with use of high-spatial-resolution MR imaging high-spatial-resolution MR imaging with histopathologic comparison. with histopathologic comparison. Radiology 2003; 227:371-377.Radiology 2003; 227:371-377.

Kim JH, Beets GL, Kim, MJ, et al. High Kim JH, Beets GL, Kim, MJ, et al. High resolution MR imaging for nodal resolution MR imaging for nodal staging in rectal cancer: are there any staging in rectal cancer: are there any criteria in addition to the size? EJR criteria in addition to the size? EJR 2004; 52:78-83.2004; 52:78-83.

Page 28: HARVARD  MEDICAL SCHOOL

ReferencesReferences Lahaye MJ, Engelen SME, Kessels AGH, Lahaye MJ, Engelen SME, Kessels AGH,

et al. USPIO-enhanced MR Imaging for et al. USPIO-enhanced MR Imaging for Nodal Staging in Patients with Primary Nodal Staging in Patients with Primary Rectal Cancer: Predictive Criteria. Rectal Cancer: Predictive Criteria. Radiology 2008; 246(3), 804-811. Radiology 2008; 246(3), 804-811.

Schnall MD, Furth EE, Rosato EF, Schnall MD, Furth EE, Rosato EF, Kressel HY. Rectal tumor stage: Kressel HY. Rectal tumor stage: Correlation of endorectal MR imaging Correlation of endorectal MR imaging and pathologic findings. Radiology and pathologic findings. Radiology 1994; 190:709-714.1994; 190:709-714.

Sauer R, Becker H, Hohenberger W, et Sauer R, Becker H, Hohenberger W, et al. Preoperative versus postoperative al. Preoperative versus postoperative chemoradiotherapy for rectal cancer. chemoradiotherapy for rectal cancer. NEJM 2004; 351;17:1731-40.NEJM 2004; 351;17:1731-40.

Page 29: HARVARD  MEDICAL SCHOOL

Secondary AnalysisSecondary AnalysisSensitivitySensitivity SpecificitySpecificity

All nodes All nodes considered considered malignantmalignant

11 00

Nodes Nodes considered considered malignant if > 3 malignant if > 3 mmmm

0.300.30 0.940.94

Nodes Nodes considered considered malignant if > 3 malignant if > 3 mmmm

0.860.86 0.510.51

Nodes Nodes considered considered malignant if > 3 malignant if > 3 mmmm

0.570.57 0.820.82

No nodes No nodes considered considered malignantmalignant

00 11

* Schnall et al (1994) and Brown et al (2003)