Drs. Roy, Subramanian and Backman are co-founders of Nanocytomics LLC and managed by the conflict of interest committees of Boston University and Northwestern University. This does not alter the authors' adherence to PLOS ONE policies on sharing data and materials.
Conceived and designed the experiments: HR CB HS KK BH VB. Performed the experiments: HS DZ CM JC LB KK BH MQ JP MP. Analyzed the data: HS DZ CM JC LB CW. Contributed reagents/materials/analysis tools: HS DZ CM JC LB CW KK. Wrote the paper: HR CB HS VB.
‡ These authors contributed equally to this work.
To determine whether nano-architectural interrogation of prostate field carcinogenesis can be used to predict prognosis in patients with early stage (Gleason 6) prostate cancer (PCa), which is mostly indolent but frequently unnecessarily treated.
We previously developed partial wave spectroscopic microscopy (PWS) that enables quantification of the nanoscale intracellular architecture (20–200nm length scale) with remarkable accuracy. We adapted this technique to assess prostate needle core biopsies in a case control study from men with Gleason 6 disease who either progressed (n = 20) or remained indolent (n = 18) over a ~3 year follow up. We measured the parameter disorder strength (Ld) characterizing the spatial heterogeneity of the nanoscale cellular structure and nuclear morphology from the microscopically normal mucosa ~150 histologically normal epithelial cells.
There was a profound increase in nano-architectural disorder between progressors and non-progressors. Indeed, the Ld from future progressors was dramatically increased when compared to future non-progressors (1±0.065 versus 1.30±0.0614, respectively p = 0.002). The area under the receiver operator characteristic curve (AUC) was 0.79, yielding a sensitivity of 88% and specificity of 72% for discriminating between progressors and non-progressors. This was not confounded by demographic factors (age, smoking status, race, obesity), thus supporting the robustness of the approach.
We demonstrate, for the first time, that nano-architectural alterations occur in prostate cancer field carcinogenesis and can be exploited to predict prognosis of early stage PCa. This approach has promise in addressing the clinically vexing dilemma of management of Gleason 6 PCa and may provide a paradigm for dealing with the larger issue of cancer overdiagnosis.
Overdiagnosis is an emerging deterrent towards implementation of cancer population screening[
One emerging approach for risk assessment is through exploiting field carcinogenesis. Field carcinogenesis (a.k.a. field cancerization, etiological field effect, field of injury etc.) is a well-established phenomenon in cancer biology that a focal neoplastic lesion develops in a permissive mutational environment from the interactions between genetic substrate and exogenous factors (diet, obesity, smoking, diabetes etc.) [
Our group has developed a novel optics technology, partial wave spectroscopic microscopy (PWS), or nanocytology, which allows quantification of cellular structure at the nanometer length scales.[
This study was approved by the Institutional Review Board at NorthShore University HealthSystem. Samples were obtained from the NorthShore University active surveillance trial initiated in November 2008. Informed written consent was obtained from the participants. The inclusion criteria for the cohort of patients included: 1) age ≥60; 2) clinical stage ≤T2a; 3) Gleason score (grade) <7; 4) ≤3/12 positive cores on initial biopsy; and 5) < 50% linear involvement of any single core. Patients underwent their first surveillance biopsy 6–12 months after enrollment. Progression was defined as any change in criteria 3, 4 or 5. Thirty-eight patients were randomly chosen from the database of patients adjudicated as progressors and non-progressors by the chief study urologist (CBB).
Transrectal biopsies were obtained under three dimensional (3-D) ultrasound guidance. Four of these cores were designated as research based and were fixed in an alcohol-based fixative and paraffin-embedded. Four micrometer sections were cut onto charged glass slides and deparaffinized for PWS. Hematoxylin and eosin (H&E) sections were reviewed by the study pathologist (MP) to direct the PWS analysis towards non-malignant tissue.
Briefly, as previously described, PWS marker disorder strength (Ld) is proportional to the mean and standard deviation of the spatial variations of the macromolecular density of the fundamental cellular building blocks (proteins, nucleic acids, lipids). Thus, Ld, colloquially, can be described as measuring the “clumpiness” of nanoscale structure. We have demonstrated that PWS is sensitive to structures from 20–200 nm through the spectral analysis of the interference spectra of light reflected from intracellular refractive index variations within microscopic spatio-temporal coherence volume, as opposed to typical light microscopy whose resolution is restricted 200–500 nm, the diffraction limit of light. Thus, histology is insensitive to many structures known to be involved in early carcinogenesis (macromolecular complexes, ribosomes, mitochondria, high-order chromatin etc.) that may be PWS detectable. Previously, we have shown that PWS was able to predict risk of colon, lung, esophagus, pancreas and ovarian cancers[
PWS analysis was performed as previously described. Regions of interest were identified (histologically normal glands) by the pathologist and calculated over ~30 regions per patient slide examined (~50K-75K pixels or ~150 cells). The operator did not have clinical information for any of the 38 samples. However, the clinical information was available to technological investigators in the first 18 (“unblinded”) but not the last 20 (“blinded”) samples was performed in the first 18 patients but not the last 20.
All statistical analyses were performed using SAS 9.3 (SAS Inc.,
Cary, NC). Patient demographics were compared between progressors and non-progressors using t-test (for continuous variables) and Chi-square test. A standard two tailed t-test (assuming unequal variances) and analysis of covariance (ANCOVA) were performed to compare the Ld between progressors and non-progressors with a p value ≤0.05 considered statistically significant. The effect size was calculated using Cohen’s
We enrolled two sets of patients, an initial cohort of 18 patients (8 non-progressors and 10 progressors) where the PWS operator was blinded to clinical status but, followed by a distinct group of 20 patients (10 progressor and 10 non-progressors) in which our PWS analysis was performed by investigators blinded to group. The demographic factors are listed in
PWS was performed in histologically normal glands identified by our pathologist. As can be noted from bright field imaging, we primarily targeted epithelium adjacent to glands for the defined region of interest. While there was no apparent difference on the bright field, there was a profound increase in Ld (more red in pseudocolor map) in progressors versus non-progressors (
The bright field images from non-progressors (a) and progressors (b) are indistinguishable on the initial surveillance biopsies. However, the PWS images in the progressors show markedly higher Ld (e.g. red areas) when compared to the non-progressors, indicating profound nano-architectural abnormalities in the pathologically normal prostate epithelium. Median length of follow up was 34.2 months.
We calculated the mean Ld from the region of interest per patient. Elevated Ld has been a hallmark of neoplastic transformation in a number of organs[
The median follow up time was 34.2 months.
This translated into a sensitivity = 88% and specificity = 72% for discriminating between progressors and non-progressors.
Since clinical factors (e.g. age, race[
The clinical need to differentiate aggressive from indolent PCa is clear since it is estimated that 233,000 American men are diagnosed annually with PCa, but only 29,480 are projected to die (still the second leading cause of male cancer deaths)[
The issue of overdiagnosis now presents a major impediment to cancer screening in other cancers as well, including mammography for breast cancer and low-dose computerized tomography (LDCT) scanning for lung cancer[
Our approach of exploiting field carcinogenesis is particularly apropos for PCa given its multifocal nature. Indeed, the Gleason score is calculated by assessing the grade of the most common and next most common (or most aggressive) clone. While assessing individual tumor clones (e.g. 3-gene immunohistochemical signature) may correlate with the natural history of some PCa[
Our previous work on nanocytological field carcinogenesis has been to assess presence of neoplasia; thus, this report represents a major new application for this approach, i.e. prognostication. Intuitively, it does seem logical to believe that field carcinogenesis profiling may be more promising than focusing on the tumor alone because of the multi-clonality that is the hallmark of prostate cancer. Indeed, the Gleason score is based on the grade of the most common and then either the next most common or most aggressive clone. Thus, it would be unclear which clone to analyze or how to synthesize gene expression data from a variety of clones. Our work parallels some of the data with liver cancer (another frequently multifocal disease) that profiling the uninvolved hepatocytes predicted recurrence in the liver after resection. [
Biologically, increased cellular nano-architectural disorder is intimately related to molecular events. From a physical perspective, an increased Ld implies “clumpiness” at the 20–200 nm length scales. These length scales encompass structures ranging from macromolecular complexes to small organelles. In the nucleus, disorder strength (Ld) is a measure of high order chromatin organization, which impacts upon multiple processes controlling gene expression.,, In keeping this this, we have demonstrated that nuclear Ld correlates with transcriptional activity[
There are many strengths of this study including its innovative nature and the well-characterized, prospectively followed cohort of men undergoing active surveillance. The clinical novelty includes use of field carcinogenesis to predict natural history of early stage PCa. From a biological perspective, this provides important insights into nano-architectural abnormalities in early carcinogenesis. Technologically, this study not only used PWS to assess prognosis, rather than just the presence of tumor, but also overcame the challenge of performing nanocytology on whole tissue sections.
Weaknesses that need to be acknowledged include modest cohort size and follow-up. On the other hand, this study represents the gold standard with the samples collected in a PRoBe-like fashion (samples prospectively collected and last half of dataset blinded; given that even in the unblinded dataset the investigator responsible for data acquisition was unaware of the clinical status, there is no possibility of bias). Another potential weakness of this study was the lack of clinically unequivocal endpoints, e.g. metastasis or death from PCa. Both of these endpoints, however, would require at least 10 years of follow up and would therefore be impractical. Furthermore, our definition of progression (increased Gleason score or increased volume of cancer) is well validated and widely used in clinical practice.
In conclusion, we demonstrate there are profound nano-architectural alterations in prostate cancer field carcinogenesis. Assessment of this through PWS nanocytology may represent a powerful biomarker to predict progression for men with early stage PCa. This approach may make PCa population screening more viable by mitigating the harms associated with identification of indolent disease. Moreover, this proof of principle study suggests that PWS nanocytology may potentially have a role in the clinical armamentarium against the screening overdiagnosis conundrum for a number of malignancies.
The authors thank Ms. Beth Parker for excellent support in manuscript preparation.