|
|
 |
 |
Search
Sign in
Information for
This is an AstraZeneca International website for healthcare professionals.
The site is not intended for the following audiences who should use the links below:
ProstateLine
BreastCancerSource
|
|
|
 |
 |
|
 |
 |
27 August 2008
|
Novel tool predicts clinical failure within 5 years of radical prostatectomy
|
MedWire News: Prostate cancer clinical failure within 5 years of radical prostatectomy can be predicted with a high degree of accuracy by a novel tool that combines clinicopathological variables with imaging and biomarker data, claim US investigators.
Currently, there are no personalized tools for the prediction of clinical failure among prostate cancer patients treated with radical prostatectomy, explain Michael Donovan, from Aureon Laboratories in Yonkers, New York, and colleagues.
The researchers developed an algorithm for predicting clinical failure in a training set of 373 prostate cancer patients who had undergone radical prostatectomy and had complete clinicopathologic, morphometric, molecular, and outcome information. The algorithm was then tested on a validation set consisting of a further 385 patients.
Biopsy cores were assessed using digitized hematoxylin and eosin (HE) images, while quantitative multiplex biomarker immunofluorescence was used to measure multiple antigens in a single tissue section.
In all, 9% of patients in the training set experienced clinical failure after radical prostatectomy, all but one of whom had received hormonal therapy, while the remaining patient received salvage radiotherapy. Follow-up lasted for a median of 76 months, and the overall median time to clinical failure was not reached.
Univariate analysis indicated that 12 features on HE-stained images were associated with clinical failure, along with 11 immunofluorescence features associated with androgen receptor and alpha-methylacyl coenzyme A racemase (AMACR).
Multivariate analysis led to an optimized model for predicting clinical failure that contained six variables: dominant prostatectomy Gleason grade; lymph node involvement; cytoplasm texture feature 1, relative area of lumen, and lumen mean border length on HE staining; and average androgen receptor intensity in AMACR-negative epithelial nuclei.
The model had a concordance index, sensitivity, and specificity of 0.92, 90%, and 91% for predicting clinical failure within 5 years of prostatectomy. When the model was applied to the validation cohort, in which 8% experienced clinical failure, the concordance index, sensitivity, and specificity were 0.84, 84% and 85%, respectively.
Interestingly, high androgen receptor levels predicted a shorter time to castrate prostate-specific antigen increases after androgen deprivation therapy, the team says in the Journal of Clinical Oncology.
They conclude: "We believe that the highly accurate predictions of the model will allow the development of more informed and appropriate treatment plans."
J Clin Oncol 2008; 26: 3923-3929
http://jco.ascopubs.org/cgi/content/abstract/26/24/3923
|
|
|
 |
 |
|
 |
Home
E-mail this story to a friend