Optimizing immunofluorescence with high-dynamic-range imaging to enhance PD-L1 expression evaluation for 3D pathology assessment from NSCLC tumor tissue

Assessing programmed death ligand 1 (PD-L1) expression through immunohistochemistry (IHC) is the golden standard in predicting immunotherapy response of non-small cell lung cancer (NSCLC). However, observation of heterogeneous PD-L1 distribution in tumor space is a challenge using IHC only. Meanwhile, immunofluorescence (IF) could support both planar and three-dimensional (3D) histological analyses by combining tissue optical clearing with confocal microscopy. We optimized clinical tissue preparation for the IF assay focusing on staining, imaging, and post-processing to achieve quality identical to traditional IHC assay. To overcome limited dynamic range of the fluorescence microscope’s detection system, we incorporated a high dynamic range (HDR) algorithm to restore the post imaging IF expression pattern and further 3D IF images. Following HDR processing, a noticeable improvement in the accuracy of diagnosis (85.7%) was achieved using IF images by pathologists. Moreover, 3D IF images revealed a 25% change in tumor proportion score for PD-L1 expression at various depths within tumors. We have established an optimal and reproducible process for PD-L1 IF images in NSCLC, yielding high quality data comparable to traditional IHC assays. The ability to discern accurate spatial PD-L1 distribution through 3D pathology analysis could provide more precise evaluation and prediction for immunotherapy targeting advanced NSCLC.

channels remained identical.
2. A collection of "pseudo images" was generated by preprocessing the PD-L1 channels of the original IF images by erosion and Gaussian blurring of kernel sizes 5 and 3, respectively.These pseudo images had the same SYTO channels and exposures as the corresponding original images.
3. A binary mask of the tissue area was obtained via the subsequent dilation of a kernel size of 21 and Otsu thresholding on the SYTO channel.
4. Let m be the total number of original and pseudo images.N pixel locations were sampled from the tissue area, such that n is the smallest integer greater than 255 / (m -1), and all these pixels had mean values greater than 16 in the original and pseudo images.5.A response curve was fitted by minimizing the quadratic objective suggested in the report by Debevac and Malik with singular value decomposition and the sampled pixel values in the original and pseudo images, except that the weighting function w(z) was offset by 16, and the regularization parameter λ was set to 100.
6.The irradiance of the PD-L1 channel of the underlying specimen was estimated using equation 6 in the report by Debevac and Malik, 1 with the offset weighting function and the fitted response curve obtained in step 5. 7. The "merged" PD-L1 channel was determined by linearly scaling the irradiance estimated in step 6 between the minimum and maximum pixel values of the PD-L1 channels of the original and pseudo images.8.A threshold α was determined by the 99.9 th -quantile of the merged PD-L1 channel in the non-tissue area (which had been computed in step 3), and pixels with merged PD-L1 values x greater than α were offset by (255 − )[( − )/(255 − )] 1/ , where  = 1.3.9.The merged PD-L1 channel was post-processed using contrast-limited adaptive histogram equalization on an 8-by-8 grid, 2 and the maximum histogram density was limited to 3. 10.The merged PD-L1 channel was post-processed via subtraction by its black-hat transformation and addition by white-hat transformation, using an ellipse kernel of size 3.This contrast enhancement procedure was performed twice, and between consecutive enhancements, Gaussian blurring of a kernel size of 3 was applied.
11. Finally, the merged PD-L1 channel was combined with the input SYTO channel to produce an HDR image of the specimen.Table S1.Conditions for three-dimensional histopathology imaging acquisition using an FV3000 confocal laser scanning microscope.

Table S2 . PD-L1 TPS of NSCLC specimens evaluated by two pathologists using IHC and IF images with medium-level exposure time as well as HDR processing.
* Laser intensity changed with a fixed slope with depth variation of tissue sections.Note: Cases 11 and 23 were excluded because the number of tumor cells in sections was < 100.PD-L1, programmed death ligand 1; TPS, tumor proportion score; NSCLC, non-small cell lung cancer; IHC, immunohistochemistry; IF, immunofluorescence; MED, medium-level exposure time; HDR, high dynamic range.

Table S3 . Full data of annotated tumor area (m 2 ) and PD-L1 expressed area (m 2 ) of IHC images, three original IF images, and HDR- processed IF images for 15 specimens.
PD-L1, programmed death ligand 1; IHC, immunohistochemistry; IF, immunofluorescence; HDR, high dynamic range; LOW, low-level exposure time; MED, medium-level exposure time; HIGH, high-level exposure time.