Introduction
Nephroblastoma (Wilms tumour) is the most common childhood renal malignancy. MR imaging is increasingly used for local staging [
1]. Nephroblastomas develop from embryonic kidney cells, containing varying amounts of tissue that represent different stages of renal development (blastema, stromal and epithelial components). Classic triphasic nephroblastoma consists of variable amounts of each of these three cell lines [
2]. There are two approaches in the treatment of nephroblastoma: The Children’s Oncology Group (COG) in North America advocates upfront surgery followed by chemotherapy, depending upon the histopathological result, whereas the International Society of Pediatric Oncology (Societe Internationale d’Oncology Pediatrique, or SIOP) in Europe focuses on using preoperative chemotherapy. Both approaches show equally high rates of overall survival [
2]. However histopathology findings have different implications depending on whether the tumour is resected before or after chemotherapy. In the SIOP trial, nephroblastomas with diffuse anaplasia or blastema-type tumour are classified as high-risk, whereas in the COG system, the blastema component has less prognostic significance. The COG classification separates tumour into three categories: favourable histology (no anaplasia), focal anaplasia or diffuse anaplasia [
2].
To improve care for these children, prognostic biomarkers for better risk stratification are needed to maximize survival with minimal toxicity. The ability to identify high-risk histopathological subtypes (such as blastema-predominant lesions after preoperative treatment) might enable more personalized treatment decisions in the future [
2,
3]. The addition of diffusion-weighted imaging (DWI) to the standard MRI protocol might provide information regarding subtype characterisation and treatment response beyond necrosis and volume change [
1,
4,
5].
Previous studies compared whole-tumour apparent diffusion coefficient (ADC) measurements with post-resection whole-tumour histopathological assessment [
4,
5]. These studies reported a relationship between ADC markers and stromal subtype histopathology. Unfortunately both studies found relatively low ADC values in both epithelial and blastemal subtypes. However using a single ADC variable to represent the whole tumour could plausibly obscure underlying correlations between ADC value and histopathology type. Direct correlation of single-slice ADC measurements with a corresponding histopathology slice might therefore refine the analysis and enhance the predictive value of DWI–MRI.
The purpose of this study was to show the feasibility of correlating MRI–DWI results with post-resection histopathology by comparing ADC metrics to histopathological subtypes in nephroblastomas.
Discussion
Our explorative pilot study shows that direct correlation between MR images and post-resection histopathology specimens is feasible in the majority of lesions. However half of the potential patients were excluded because of extensive cystic, haemorrhagic or necrotic changes that occurred during preoperative treatment. With this limited number of lesions, we found a strong linear relationship between the stromal proportion at histopathology and median ADC. Our results replicate previous studies in 25 patients [
4,
5]. However with side-by-side comparison of MR images and histopathology findings, the current reported linear relationship is more apparent. Likewise, there was a strong inverse linear relationship between percentage of blastema at histopathology and the lower quartile ADC. Our previous reported study found a similar, however weak, relation. Unfortunately only limited proportions of the epithelial subtype were present in our current reported study. Therefore potential linear relationship between 25th percentile ADC and percentage of blastemal components might be overestimated because both blastema and epithelial predominant lesions demonstrate relatively low ADC values [
4,
5]. Furthermore the overall effect in a group of tumours cannot be directly extrapolated to the individual patient. If there is a significant overlap in ADC values, a single ADC value might be of limited clinical value for the individual patient. Therefore, reliable differentiation between epithelial- and blastemal-predominant lesions at presentation is probably not possible with ADC measurements alone. However identification of a considerable proportion of relatively low ADC values at presentation could guide tumour biopsy. With the combination of MRI–DWI findings and tumour biopsy results at presentation, the high-risk predominant blastemal subtype nephroblastoma could potentially be identified to help guide personalized treatment decisions [
2].
The strong linear relationship between stromal histopathology and median ADC could be of additional clinical value, particularly in bilateral disease. In bilateral disease, accurate assessment of treatment response is important to direct therapy planning in order to spare as much renal function as possible. Second-line treatment chemotherapy is added or substituted when the tumour shrinkage appears poor [
8]. However, following chemotherapy stromal-predominant nephroblastomas tend to differentiate into more mature stromal or mesenchymal tumor types [
8]. Median ADC measurements could be a promising tool to identify stromal-predominant tumours that could respond to preoperative treatment with differentiation instead of shrinkage.
Previous studies have suggested that blastema is the most responsive tumour component to chemotherapy [
9,
10]. Moreover, chemo-resistant blastemal subtype after preoperative treatment has prognostic value with respect to event-free and overall survivial [
1,
2]. MRI with DWI might be of value to assess both tumour response to treatment and ADC values in residual viable parts of lesions [
5]. The combination of evident tumour shrinkage or necrosis with relative low ADC values in viable tumour components could be suggestive of chemo-resistant blastema [
4,
5]. Therefore identification of residual blastemal components with DWI might guide treatment decisions, including surgical planning of a nephron-sparing approach in bilateral disease.
Further prospective studies with a larger cohort of children should be performed to validate our preliminary findings. To achieve reliable side-by-side comparison between MRI findings and histopathology, a collaborative effort is essential. First, effective information transmission from surgeon to histopathologist during handling of the post-resection specimen is required. Furthermore, review of sliced specimen needs input from radiologist and histopathologist. In this explorative study, we have shown that with these efforts, side-by-side comparison is feasible.
Our study has several limitations. First, because of its retrospective nature, a selection bias could have been introduced, which weakens the general validity of our data. Second, an interpretation bias was induced by the match between histopathology slides and MR images. Likely there was a varying degree of difference in orientation of slicing between post-resection specimens and MRI images. However, this side-by-side comparison could be further improved by using agar fixation of the resection specimen in order to obtain thinner slices as demonstrated in Fig.
1. An important disadvantage of using agar fixation is the lack of availability of obtaining tissues for biological studies. Third, histopathological estimation of tumour composition is not perfectly objective. However the overall tumour subtype classification was concordant with the central review. Furthermore the interobserver variability of our method was tested for the whole-tumour approach. According to previous studies, a single-slice approach is subject to slightly broader reader variability [
11]. Finally, we used the enhancement of the erector spinae muscles as a threshold filter for excluding less-enhancing portions of lesions. Formal perfusion analysis could further improve the selection of viable areas of the lesions.
Prospective studies with a larger cohort of patients could further elucidate the potential additional role of DWI and ADC measurements in pretreated nephroblastoma.
Acknowledgements
We would like to thank Rutger Jan Nievelstein, MD, PhD, and Prof. Willem P. Th. M. Mali, MD, PhD, for important contributions to this study.
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.