Prediction of Prostate Cancer Recurrence Using Quantitative Phase Imaging

The risk of biochemical recurrence of prostate cancer among individuals who undergo radical prostatectomy for treatment is around 25%. Current clinical methods often fail at successfully predicting recurrence among patients at intermediate risk for recurrence. We used a label-free method, spatial light interference microscopy, to perform localized measurements of light scattering in prostatectomy tissue microarrays. We show, for the first time to our knowledge, that anisotropy of light scattering in the stroma immediately adjoining cancerous glands can be used to identify patients at higher risk for recurrence. The data show that lower value of anisotropy corresponds to a higher risk for recurrence, meaning that the stroma adjoining the glands of recurrent patients is more fractionated than in non-recurrent patients. Our method outperformed the widely accepted clinical tool CAPRA-S in the cases we interrogated irrespective of Gleason grade, prostate-specific antigen (PSA) levels and pathological tumor-node-metastasis (pTNM) stage. These results suggest that QPI shows promise in assisting pathologists to improve prediction of prostate cancer recurrence.


Gleason Grading
The Gleason grade is the most widely used grading scheme in prostate cancer. The system was developed by Dr. Donald Gleason in the 1960s and is based on glandular differentiation seen in hematoxylin and eosin (H&E) stained slides [1,2]. Pathologists determine Gleason grade based on glandular presence and differentiation in stroma [1,2]. The Gleason score has been proven to be an indicator of tumor size, metastasis, treatment and outcome [2][3][4][5]. The variation in grades on a scale from 1-5, in order of increasing severity, is based on glandular differentiation and glandular presence in stroma [1,2,6,7]. The primary grade is the pattern present in maximum biopsy area and the secondary grade is the second most prominent pattern. The two grades are added to provide a Gleason score of 1-10.
In current clinical practice, grades 1 and 2 are rarely diagnosed. Gleason grade 3 glands are medium to small size, singular with infiltrating edges as seen in Figs S1A,B which shows a side-by-side comparison of H&E and SLIM images. In a zoomed-in version of Gleason grade 3 in figure S2A, stroma between adjacent glands is clearly visible, thus, aiding the determination of the grade [2]. Gleason grade 4 consists of small glands that are fusing into one another (Figs S1C, D; Fig S2B) [2]. Gleason grade 5 represents the most severe cancer where glandular architecture is lost and epithelial cells are distributed individually or in sheet like patterns in the stroma (Figs S1E, F) [1]. In the zoomed-in version in Fig S2C, no glands are visible and epithelial cells can be seen between stromal fibers.

D'Amico Risk Classification
The D'Amico risk classification is based on the combination of clinical parameters such as prostate specific antigen (PSA) levels measured in blood, Gleason score on the biopsy and prostate tumor size (T) as measured in either a digital rectal exam or trans-rectal ultrasound. There are three categories [8]: 1. D'Amico low risk category: Blood PSA level ≤ 10 ng/ml, Gleason score ≤ 6, T1-T2a. (T1: Tumor was an incidental finding that is not palpable; T2a: Tumor is in less than one half of one side of the prostate) 2. D'Amico intermediate risk classification: Blood PSA level of 10-20 ng/ml, Gleason score 7, T2b. (T2b: Tumor is confined to one side of the prostate, but more than one half of one lobe) 3. D'Amico high risk classification: Blood PSA level > 20 ng/ml, Gleason score ≥ 8, T2c-T3a. (T2c: Tumor is in both sides of the prostate; T3a: Tumor shows extra-capsular extension) As displayed in the profile distribution histograms in Figure S3, the majority of the cases used in our study are, by design, from the D'Amico intermediate risk category.

CAPRA-S Score
The Cancer of the Prostate Risk Assessment (CAPRA-S) score is a commonly used post-radical prostatectomy prostate cancer recurrence risk assessment tool, which is described in detail elsewhere [9]. It assigns differential weightage to PSA levels before surgery, pathological Gleason score, extra-capsular extension, surgical margin status, seminal vesicle invasion and lymph node invasion. The weightages are added up to provide a score that determines the risk of biochemical recurrence of prostate cancer.

Receiver Operating Characteristic (ROC)
The ROC curve plots the sensitivity (true positive rate) against 1-specificity (false positive rate). In order to plot an ROC curve using binary data, incremental cut-off or threshold values are set at which the true positive and false positive rates are determined and plotted. The area under the ROC curve (AUC) (this is equivalent to c-index in a binary outcome, however, c-index is calculated differently) represents the accuracy of the classification. An AUC of 1 corresponds to a perfect classification method whereas an AUC of 0.5 corresponds to random guess, like a coin toss.

Anisotropy factor, g
The anisotropy factor is the average cosine of the scattering angle, Using the scattering-phase theorem, we calculated g across a specimen in terms of the gradient and variance of the spatial phase distribution (Eq. 1 in text). The anisotropy values are averaged across the stroma adjoining multiple glands from 3-4 cores per patient to obtain the final g-value for each patient. Figure S4 illustrates the low-and high-values of g: more isotropic scattering corresponds to lower values of g. Our results indicate that bad outcomes are correlated to lower g-values. The morphology of stroma adjoining glands among patients with high PSA levels (PSA>20ng/ml), is fractionated and shows non-uniform swelling. This causes anisotropy to fail at identifying non-recurrent individuals with high pre-surgical PSA levels as illustrated in Figure S5.