Prediction of prostate cancer progression by analysis of selected predictive parameters

• US 6,025,128 A
• Filed: 09/29/1994
• Issued: 02/15/2000
• Est. Priority Date: 09/29/1994
• Status: Expired due to Term

• Current Assignee: Cytodiagnostics, Inc., Johns Hopkins University, Laboratory Corporation Of America Holdings, University of Tulsa
• Original Assignee: Cytodiagnostics, Inc., Johns Hopkins University, University of Tulsa
• Inventors: Bacus, Michael P., Miller, M. Craig, Epstein, Jonathan I., Coffey, Donald P., Veltri, Robert W, Ashenayi, Kaveh, Partin, Alan W.
• Primary Examiner(s): EYLER, YVONNE L
• Application Number:
  US08/315,210
• Time in Patent Office: 1,965 Days
• Field of Search: 435/7.23, 435/6, 436/63, 436/64, 436/813
• US Class Current: 435/6.14
• CPC Class Codes: C12Q 1/68 involving nucleic acids
• CPC Class Codes: G01N 33/574 for cancer
• CPC Class Codes: G01N 33/57434 of prostate
• CPC Class Codes: G16B 20/00 ICT specially adapted for functional genomics or proteomics, e.g. genotype-phenotype associations
• CPC Class Codes: G16B 25/10 Gene or protein expression profiling; Expression-ratio estimation or normalisation
• CPC Class Codes: G16B 40/00 ICT specially adapted for biostatistics; ICT specially adapted for bioinformatics-related machine learning or data mining, e.g. knowledge discovery or pattern finding
• CPC Class Codes: G16B 40/20 Supervised data analysis
• CPC Class Codes: G16H 50/20 for computer-aided diagnosis, e.g. based on medical expert systems
• CPC Class Codes: G16H 50/50 for simulation or modelling of medical disorders
• CPC Class Codes: Y10S 436/813 Cancer

Abstract

A method for screening individuals at risk for prostate cancer progression is disclosed. The method is useful for evaluating cells from patients at risk for recurrence of prostate cancer following surgery for prostate cancer. Specifically, the method uses specific Markovian nuclear texture factors, alone or in combination with other biomarkers, to determine whether the cancer will progress or lose organ confinement. In addition, methods of predicting the development of fatal metastatic disease by statistical analysis of selected biomarkers is also disclosed. The invention also contemplates a method that uses a neural network to analyze and interpret cell morphology data. Utilizing Markovian factors and other biomarkers as parameters, the network is first trained with a sets of cell data from known progressors and known non-progressors. The trained network is then used to predict prostate cancer progression in patient samples.

Claims

1. A method of predicting prostate cancer progression, comprising:

(a) obtaining prostate cells from a subject;

(b) analyzing predictive parameters in the prostate cells, wherein the predictive parameters are nuclear morphometric descriptors, including;

object sum optical density, picograms of DNA, contrast, correlation, sum average, sum variance, difference variance, difference entropy, information measure B, product moment, standard deviation, and DNA ploidy; and

(c) predicting cancer progression by statistical analysis of the predictive parameters, where the statistical analysis is logistic regression, discriminate analysis, recursive partitioning, neural network, or classification and regression tree analysis.

2. The method of claim 1, wherein the nuclear morphometric descriptors are selected from the group consisting of object sum optical density, object shape, picograms of DNA, contrast, correlation, inverse difference moment, sum average, sum variance, difference variance, difference entropy, information measure B, product moment, standard deviation, and DNA ploidy.

3. The method of claim 1, wherein the nuclear morphometric descriptors are selected from the group consisting of object sum optical density, object size, object shape, picograms of DNA, angular second moment, contrast, correlation, difference moment, inverse difference moment, sum average, sum variance, sum entropy, entropy, difference variance, difference entropy, information measure A, information measure B, maximal correlation coefficient, coefficient of variation, peak transition probability, diagonal variance, diagonal moment, second diagonal moment, product moment, triangular symmetry, blobness, standard deviation, and DNA ploidy.

4. The method of claim 1, wherein the predictive parameters further include Post-op Gleason.

5. The method of claim 4, wherein the predictive parameters further include PD-41 antigenic expression and Her-2-neu antigenicity.

6. The method of claim 1, wherein the nuclear morphometric descriptors are Markovian nuclear texture features.

7. The method of claim 1, wherein the statistical analysis is univariate or multivariate analysis.

8. The method of claim 7, wherein the statistical analysis is multivariate analysis.

25. The method of claim 1, 9 or 13, further comprising analyzing one or more nuclear morphometric descriptors selected from the group consisting of object size, object shape, angular second moment, difference moment, inverse difference moment, sum entropy, entropy, information measure A, maximal correlation coefficient, coefficient of variation, peak transition probability, diagonal variance, diagonal moment, second diagonal moment, triangular symmetry, perimeter, DNA index, density, average optical density, feret X, feret Y, maximum diameter, minimum diameter and elongation.

9. A method for predicting the recurrence of prostate cancer following radical prostatectomy comprising the steps of:

(a) obtaining prostate cells from a subject;

(b) generating nuclear morphometric descriptors for the cells, including;

object sum optical density, picograms of DNA, contrast, correlation, sum average, sum variance, difference variance, difference entropy, information measure B, product moment, standard deviation, and DNA ploidy; and

(c) predicting the recurrence of prostate cancer in the cell samples by statistical analysis of the nuclear morphometric descriptors, where the statistical analysis is logistic regression, discriminate analysis, recursive partitioning, neural network, or classification and regression tree analysis.

10. The method of claim 9, wherein the statistical analysis is multivariate statistical analysis.

11. The method of claim 10, further comprising statistical analysis of predictive parameters selected from the group consisting of Post-op Gleason, nuclear roundness variance, PD-41 antigenic expression and Her-2-neu antigenic expression.

12. The method of claim 11, wherein the nuclear morphometric descriptors are Markovian nuclear texture features.

13. A method of predicting the occurrence of fatal metastatic prostate disease comprising the steps of:

(a) obtaining prostate cells from a subject;

(b) generating nuclear morphometric descriptors for the cells, including;

object sumn optical density, picograms of DNA, contrast, correlation, sum average, sum variance, difference variance, difference entropy, information measure B, product moment, and standard deviation; and

(c) predicting the occurrence of fatal metastatic prostate disease by statistical analysis of the nuclear morphometric descriptors.

14. The method of claim 13, further comprising statistical analysis of predictive parameters selected from the group consisting of post-op Gleason DNA ploidy, PD-41 antigenic expression, or Her-2-neu antigenic expression.

15. The method of claim 14, wherein the predictive parameter is Her-2-neu antigenic expression.

30. The method of claim 13 or claim 16, wherein the statistical analysis is logistic regression, discriminate analysis, recursive partitioning, classification and regression tree analysis, or neural network.

16. A method of predicting the progression of prostate cells from a normal state to a malignant state comprising the steps of:

(a) obtaining prostate cells from a subject;

(b) generating nuclear morphometric descriptors for the cells;

(c) analyzing selected cell biomarkers; and

(d) predicting the progression of the cells by using multivariate statistical modeling of the nuclear morphometric descriptors and the selected biomarkers.

17. The method of claim 16, wherein the selected biomarkers are PD-41 antigenic expression and Her-2-neu antigenic expression.

18. The method of claim 16, wherein the nuclear morphometric descriptors include Markovian texture features.

26. The method of claim 16 or claim 19, wherein the nuclear morphometric descriptors are selected from the group consisting of object sum optical density, object size, object shape, picoprams of DNA, angular second moment, contrast, correlation, difference moment, inverse difference moment, sum average, sum variance, sum entropy, entropy, difference variance, difference entropy, information measure A, information measure B, maximal correlation coefficient, coefficient of variation, peak transition probability, diagonal variance, diagonal moment, second diagonal moment, product moment, triangular symmetry, standard deviation, DNA ploidy, perimeter, DNA index, density, average optical density, feret X, feret Y, maximum diameter, minimum diameter and elongation.

31. The method according to claim 17, wherein the patient derived pathology and clinical information variables are selected from the group consisting of post operative Gleason score, serum PSA, PD-41 antigenic expression, or Her-2-neu antigenic expression.

19. A method of determining prostate cancer progression comprising:

(a) providing a neural network;

(b) training the neural network using predictive parameters, obtained from prostate cells known to progress and a set of predictive parameters obtained from prostate cells known not to progress, the predictive parameters comprising nuclear morphometric descriptors;

(c) analyzing dredictive parameters in tumor cells of an individual having an unknown state of cancer progression; and

(d) predicting cancer progression in cells of the individual having an unknown state of cancer progression using the predictive parameters and the trained neural network.

20. The method of claim 19, wherein the nuclear morphometric descriptors are selected from the group consisting of object sum optical density, object shape, picograms of DNA, contrast, correlation, inverse difference moment, sum average, sum variance, difference variance, difference entropy, information measure B, product moment, standard deviation, and DNA ploidy.

21. The method of claim 19, wherein the nuclear morphometric descriptors are selected from the group consisting of object sum optical density, object size, object shape, picograms of DNA, angular second moment, contrast, correlation, difference moment, inverse difference moment, sum average, sum variance, sum entropy, entropy, difference variance, difference entropy, information measure A, information measure B, maximal correlation coefficient, coefficient of variation, peak transition probability, diagonal variance, diagonal moment, second diagonal moment, product moment, triangular symmetry, blobness, standard deviation, and DNA ploidy.

22. The method of claim 19, wherein the neural network is of the back propagation type.

23. The method of claim 19, wherein the neural network is of a hybrid type.

24. The method of claim 19, wherein the prognostic parameters of step (b) further include post operative Gleason score, PD-41 antigenic expression, or Her-2-neu antigenic expression.

27. A method of predicting the progression of prostate cancer comprising the steps of:

(a) obtaining prostate cells from a subject;

(b) analyzing predictive parameters from the prostate cells, the predictive parameters including nuclear morphometric descriptors;

(c) utilizing statistical analysis to determine multivariately significant nuclear morphometric descriptors to calculate a quantitative nuclear grade; and

(d) predicting the probability of progression in the patient by statistical analysis of the quantitative nuclear grade.

28. The method of claim 27, further including the steps of analyzing predictive parameters from the prostate cells including utilizing statistical analysis to determine univariately significant patient derived pathology and clinical information variables that contribute to a multivariate model solution and predicting the probability of progression in the patient by further statistical analysis of the quantitative nuclear grade and urivariately significant patient derived pathology and clinical information variables.

29. The method of claim 27, wherein the nuclear morphometric descriptors include Markovian nuclear texture features.