1.

INTRODUCTION

98.9% of men diagnosed with prostate cancer between the years 2004 - 2010 survived 5 years or longer ¹. When the data is further categorized based on disease stage at diagnosis, 80.7% of the cancers were diagnosed at the local stage, during which the 5 year survival rate is 100% ¹. However, the 5-year survival rate for individuals with distant metastasis of prostate cancer is 28%. Hence diagnosis of prostate cancer at an early stage, when it has a higher chance for cure, is critical for patient survival. However, various studies have shown that not all prostate cancers diagnosed at an early stage require intervention ^2,3. A particularly compelling argument towards this end was provided from a study by Porter et. al. that showed that 1.2 million new cases of prostate cancer would be diagnosed if all men between the ages of 62 -75 years old underwent prostate biopsies, irrespective of their PSA levels ⁴. Various nomograms and risk assessment tools have been developed to identify men with prostate cancer who have a higher risk of disease progression and mortality. The most commonly used tools include D’Amico risk stratification, CAPRA score, Kattan nomogram and Partin tables ^5-8. A patient’s Gleason grade is an important consideration in each of these tools.

A patient’s Gleason grade is determined by the pathologist based on the level of glandular differentiation seen in the H&E stained biopsy. A primary and secondary Gleason grade (1-5) is assigned based on the amount of biopsy area containing a particular pattern. The Gleason grades are added up to assign a Gleason score (1-10). The Gleason grade is considered to be an indicator of possible tumor size, metastasis and outcome ^9-12. A Gleason score of 6 or less is a common criteria used to adopt active surveillance as a prostate cancer management strategy ¹³. Thus it’s critical to ensure that the Gleason grade is correctly determined. The qualitative nature of Gleason grading can lead to a high degree of disagreements among various pathologists’ diagnoses. A study by Allsbrook et. al. showed that the level of under-grading of Gleason score 7 prostate cancer was 47% and that for Gleason score 8-10 prostate cancers was 25% among general pathologists ¹⁴. When the same study was repeated on urologic pathologists, the agreement rate was 70% or higher in the Gl¹⁵eason score ranges of 2-4, 5-6, 7, 8-10 among the 38 consensus cases but there were 10 nonconsensus cases ¹⁶ Clearly, a quantitative technique that can objectively predict Gleason grades with a high degree of accuracy would assist pathologists in performing Gleason grading. This would also have important implications on prostate cancer treatment.

Quantitative phase imaging (QPI) ¹⁷of unstained tissue provides information on the refractive index distribution, or tissue morphology, with nanometer level sensitivity. Subtle morphological changes in both the epithelial and stromal regions of tissue, which are not visible in stained tissue sections used in current pathological settings, can be measured using QPI^15,18-25Thus qpi would be a valuable addition to current diagnostic pathology. In this paper, we show the ability of QPI to help pathologists quantitatively and therefore, objectively differentiate between Gleason grade 3 and 4 prostate cancers.

2.

EXPERIMENTAL DETAILS

2.2

SLIM Imaging System

Spatial Light Interference Microscopy (SLIM) is a quantitative phase imaging system that was developed as an add-on module to a commercial phase contrast microscope and is described in detail in Ref. ¹⁵. Briefly, the back focal plane of the phase contrast objective is projected onto a liquid crystal phase modulator (LCPM). At the LCPM, three π/2 phase shifts, additional to the one already present in the phase contrast image, are introduced and recorded by the CCD. The four intensity images corresponding to each phase shift are recorded, and a final phase image is computed from the information. The final phase image computed corresponds to:

Where ϕ(x,y) is the phase at a given point, λ is the center wavelength of the white light source (552.3nm), n(x,y,z) is the refractive index at a given point and n₀ is the refractive index of the surrounding medium. The refractive index difference is integrated over the entire thickness of the sample to measure the phase. SLIM has a transverse resolution of 0.4 microns and a spatial path length sensitivity of 0.3nm.

The SLIM system was modified to raster scan through large fields of view and stitch the frames together. This provides the high throughput necessary for whole slide imaging. The samples used in this study were imaged using the 40X/0.75NA objective of the SLIM system.

2.3

Optical Anisotropy

Optical anisotropy is defined as the average cosine of the scattering angle associated with a single scattering event. Since QPI records both the amplitude and the phase of the light passing through the sample, it has the ability to measure light scattering parameters. Using the scattering phase theorem, the optical anisotropy of light passing through the tissue can be measured as ²⁸

Where k0 is the wave number of the light source, ∇[φ(r)] is the phase gradient and Δφ²(r) is the variance of the phase. The phase gradient and phase variance are averaged over a tissue region of interest, r. In our study, the region of interest was a single layer of stroma immediately adjoining the glands, as shown in Fig. 1.

Figure 1:

Stromal Region Selection. A single layer of stroma immediately adjoining these glands is highlighted in red in the above image. We calculate optical anisotropy in this layer of stroma using the scattering phase theorem.

2.4

Solidity

Glandular solidity is defined the ratio of the area of the gland to that of the area of a convex hull fit around the gland (Fig. 2). This parameter is an indicator of the uniformity of the edges of the selected tissue region. In order to measure glandular solidity, the glands of interest were selected using the ROI feature on ImageJ. ImageJ was then used to measure the area of the gland, fit a convex hull around the gland, measurement of the area of the convex hull and finally, the solidity.

Figure 2:

Glandular Solidity. Solidity is defined as the ratio of the area of the gland to the area of the convex hull or polygonal fit around the gland. In the above image, the inner boundary is the ROI of the gland, and the outer boundary shows the convex hull around the gland.

3.

RESULTS

43 cases from the TMA 2 data set with consensus diagnosis from 3 pathologists were analyzed. 22 cores with only Gleason grade 3 glands from patients with a final diagnosis of Gleason 3+3 cancer and 21 cores with only Gleason 4 glands and final diagnosis of Gleason 4+4 cancer were imaged using SLIM. Solidity, which the ratio of the area of the gland to that of a convex hull fit around the gland, was measured for all glands in the cores. Additionally, anisotropy, which is the average cosine of the scattering angle, was measured for the layer of stroma surrounding all the glands. Our study showed that Gleason 3 glands had a higher degree of solidity and higher values of anisotropy in the stroma immediately adjoining the gland (Fig. 3). Using glandular solidity and stromal anisotropy alone, 81% of the Gleason grade 4 cores and 82% of the Gleason grade 3 cores were correctly classified. We believe the higher glandular solidity values can be explained by the reduced invasive edges and lower degree of glandular fusion seen in Gleason 3 glands. Additionally, the lower value of anisotropy seen around Gleason grade 4 glands means the stromal layer has a higher level of disorder in advanced stages of disease.

Figure 3

Diagnosis of Gleason Grades 3 and 4 Gleason grade 3 glands have a higher degree of solidity and the stroma surrounding Gleason grade 3 glands have a higher value of anisotropy. Glandular solidity and stromal anisotropy can be used to identify Gleason grade 3 glands with 82% accuracy and Gleason grade 4 glands with 81% accuracy.

4.

CONCLUSION AND FUTURE WORK

QPI was able to distinguish between the Gleason grade 3 and 4 prostate cancers with an accuracy of 81.39%. QPI can thus be used to assist pathologists with Gleason grading, especially in difficult cases. Further studies need to be conducted on biopsies to determine if the parameters measured using QPI can be used to predict disease progression with a higher degree of accuracy than existing methods.