FOXP3-stained image analysis for follicular lymphoma: Optimal adaptive thresholding with maximal nucleus coverage

[Follicular lymphoma: first - line selection criteria of treatment]

Follicular lymphoma (FL) is a tumor that develops from the B cells of the germinal center; characterized by recurrent and remitting course of the disease, the transformation of a tumor into diffuse large B-cell lymphoma (DLBCL) is possible. In generalized lesions and progression of FL, the most commonly used courses are R-CHOP and R-B. The choice of therapy for different cytological types, clinical and laboratory parameters remains disputable.

AIM

To analyze the clinical, laboratory, morphological parameters of patients with FL, who got R-B and R-CHOP therapy; determine the criteria for selecting induction therapy.

MATERIALS AND METHODS

The study included 203 patients with FL from 2000 to 2018. R-CHOP treatment was initiated in 126 patients, 14 of whom later received high - dose therapy (HDT) (R-DHAP: rituximab, dexamethasone, cisplatin, cytarabine) without autologous stem cell transplantation (autoSCT), 21 - HDT with autoSCT; treatment of 89 patients was limited to courses of R-CHOP and maintenance therapy with rituximab, two patients (in whom the disease progressed, despite R-CHOP therapy) were assigned the mNHL-BFM-90 program. The efficacy of treatment on various treatment regimens was evaluated primarily by overall survival.

RESULTS AND DISCUSSION

R-B. 77 patients received R-B therapy. Complete remission of the disease was achieved in 47/77 (61%) patients (3 of them later developed a relapse of the disease), partial remission was achieved in 15/77 (19%) patients, in 13/77 (17%) cases progression was recorded tumors. 70 patients had 1-2 cytological type of tumor, 6 patients - 3A cytological type. In cases of progression, 3 of 13 patients (46%) were diagnosed with 3A cytological type FL. Median observation (at the time of analysis) - 34 months. R-CHOP. 89 patients with FL received high - dose therapy with R-CHOP (6-8 courses) and maintenance therapy with rituximab. In 39 (44%) patients, the disease remained in remission, and in 50 (56%), a relapse of the disease developed. 50 patients had 1-2 cytological types, 39 - 3 cytological types. In cases of recurrence of FL, a 3A cytologic type (36%) was diagnosed in 18/50 patients. Median observation - 93 months. R-CHOP + HDT and autoSCT. 21 patients after the R-CHOP courses continued (due to insufficient antitumor response) high - dose chemotherapy (HDT) and auto-SCT were performed. In 18/21 (86%) cases, complete remission of the disease was achieved and maintained, in 3 (14%) cases relapse developed. 16 patients had 1-2 cytological types, 5 - 3 cytological types. Median observation - 81 months. R-CHOP + HDT without autoSCT. 14 patients started therapy under the R-CHOP program as induction therapy, but then (due to insufficient antitumor response), the treatment was continued according to the HDT without autoSCT. 11 (79%) patients are currently in remission of the disease, in 3 (21%) - there was a relapse. 10 patients had 2 cytological types of PL, 4 - 3 cytological types. 11 (79%) patients are currently in remission of the disease, in 3 (21%) - there was a relapse. Median observation - 80 months. 7-year OS of patients with FL on RB therapy was 89% (95% CI 75-99), on R-CHOP therapy - 85% (95% CI 73-90), on R-CHOP + HDT and autoSCT - 87% (95% CI 57-100), on R-CHOP + HDT without autoSCT - 82%. 7-year PFS of FL patients on RB therapy was 70% (95% CI 75-99), on R-CHOP therapy - 44% (95% CI 73-90), on R-CHOP + HDT and autoSCT - 74% (95% CI 57-100), on R-CHOP + HDT without autoSCT - 80%.

CONCLUSION

The R-B is most effective in FL 1 and 2 cytological types. The cytological type does not correspond to the type of tumor growth: at 3A and 3A + 3B cytological types, nodular / nodular - diffuse and diffuse types of growth are found. When choosing an induction course, one should look at the cytological type of FL. A high proliferative activity index (according to Ki67) is a predictor of resistance to R-B therapy. The absence of an interfollicular T-cell reaction in tumor tissue FL is associated with tumor chemoresistance. The presence of the bulky factor is associated (in most patients) with the FLIPI index with values from 3 to 5, and is a predictor of a poor response to therapy. Patients with bulky, high (more than 35%) Ki67 index and FLIPI from 3 to 5 in the debut of the disease as the first line therapy, it is preferable to choose the R-CHOP mode, and in the absence of (after 4-6 courses) to complete or partial remission to continue conducting the HDT.

DeepFocus: Detection of out-of-focus regions in whole slide digital images using deep learning

The development of whole slide scanners has revolutionized the field of digital pathology. Unfortunately, whole slide scanners often produce images with out-of-focus/blurry areas that limit the amount of tissue available for a pathologist to make accurate diagnosis/prognosis. Moreover, these artifacts hamper the performance of computerized image analysis systems. These areas are typically identified by visual inspection, which leads to a subjective evaluation causing high intra- and inter-observer variability. Moreover, this process is both tedious, and time-consuming. The aim of this study is to develop a deep learning based software called, DeepFocus, which can automatically detect and segment blurry areas in digital whole slide images to address these problems. DeepFocus is built on TensorFlow, an open source library that exploits data flow graphs for efficient numerical computation. DeepFocus was trained by using 16 different H&E and IHC-stained slides that were systematically scanned on nine different focal planes, generating 216,000 samples with varying amounts of blurriness. When trained and tested on two independent datasets, DeepFocus resulted in an average accuracy of 93.2% (± 9.6%), which is a 23.8% improvement over an existing method. DeepFocus has the potential to be integrated with whole slide scanners to automatically re-scan problematic areas, hence improving the overall image quality for pathologists and image analysis algorithms.

Nuclear IHC enumeration: A digital phantom to evaluate the performance of automated algorithms in digital pathology

Automatic and accurate detection of positive and negative nuclei from images of immunostained tissue biopsies is critical to the success of digital pathology. The evaluation of most nuclei detection algorithms relies on manually generated ground truth prepared by pathologists, which is unfortunately time-consuming and suffers from inter-pathologist variability. In this work, we developed a digital immunohistochemistry (IHC) phantom that can be used for evaluating computer algorithms for enumeration of IHC positive cells. Our phantom development consists of two main steps, 1) extraction of the individual as well as nuclei clumps of both positive and negative nuclei from real WSI images, and 2) systematic placement of the extracted nuclei clumps on an image canvas. The resulting images are visually similar to the original tissue images. We created a set of 42 images with different concentrations of positive and negative nuclei. These images were evaluated by four board certified pathologists in the task of estimating the ratio of positive to total number of nuclei. The resulting concordance correlation coefficients (CCC) between the pathologist and the true ratio range from 0.86 to 0.95 (point estimates). The same ratio was also computed by an automated computer algorithm, which yielded a CCC value of 0.99. Reading the phantom data with known ground truth, the human readers show substantial variability and lower average performance than the computer algorithm in terms of CCC. This shows the limitation of using a human reader panel to establish a reference standard for the evaluation of computer algorithms, thereby highlighting the usefulness of the phantom developed in this work. Using our phantom images, we further developed a function that can approximate the true ratio from the area of the positive and negative nuclei, hence avoiding the need to detect individual nuclei. The predicted ratios of 10 held-out images using the function (trained on 32 images) are within ±2.68% of the true ratio. Moreover, we also report the evaluation of a computerized image analysis method on the synthetic tissue dataset.