Paradoxical Response to Neoadjuvant Therapy in Undifferentiated Pleomorphic Sarcoma: Increased Tumor Size on MRI Associated with Favorable Pathology

Mariam H. Goreish, Nicolò Gennaro, Laetitia Perronne, Görkem Durak, Amir A. Borhani, Hatice Savas, Linda C. Kelahan, Ryan Avery, Kamal Subedi, Tugce Agirlar Trabzonlu, Ulaş Bağcı, Baris Turkbey, Spyridon Bakas, Sean Sachdev, Ronen Sumagin, Borislav A. Alexiev, Pedro Hermida de Viveiros, Seth M. Pollack, Yury VelichkoCancers 2025 · DOI

TL;DR

This study reveals a counterintuitive finding in undifferentiated pleomorphic sarcoma (UPS) treatment: an increase in tumor size on MRI after neoadjuvant therapy can paradoxically indicate a favorable pathological response. This challenges standard imaging criteria like RECIST, suggesting that size changes alone are insufficient for assessing treatment efficacy in UPS.

What Problem Does This Paper Address?

Undifferentiated pleomorphic sarcoma (UPS) is an aggressive soft tissue sarcoma with a high risk of local recurrence. Neoadjuvant therapy (treatment before surgery) is used to improve outcomes, but accurately assessing its effectiveness is challenging. Current standard imaging criteria, like RECIST (Response Evaluation Criteria in Solid Tumors), primarily rely on tumor shrinkage. However, previous studies have shown that UPS tumors don't always shrink and can even increase in size after treatment, leading to uncertainty about whether this increase signifies disease progression or a positive treatment response. This paper aims to clarify this by correlating MRI-based tumor size changes with pathological response and local recurrence in UPS patients.

Methods

This was a retrospective study analyzing data from 39 patients with biopsy-proven UPS who received neoadjuvant chemoradiotherapy (nCRT). The key methodological components were:

  • Imaging Analysis: Four readers independently measured tumor dimensions (two axial diameters and one coronal/sagittal diameter) from pre- and post-treatment MRI scans. They also calculated cross-sectional areas and volume.
  • Pathological Response Classification: Patients were classified as pathological responders (pR) if they had ≤10% viable tumor cells after treatment, and non-responders (pNR) if they had >10% viable cells.
  • Statistical Analysis: Inter-reader agreement was assessed using Kendall's concordance coefficient. One-way ANOVA and Tukey's HSD test were used to compare tumor size changes between pR and pNR groups. The study also explored a fully Bayesian approach using generalized linear mixed-effects models (GLMMs) to account for inter-reader variability. Optimal cutoffs for predicting response and recurrence were determined using ROC curve analysis.

Datasets

The study utilized a dataset of 39 patients with biopsy-proven Undifferentiated Pleomorphic Sarcoma (UPS) who underwent neoadjuvant therapy. This data was sourced from an institutional electronic data warehouse. No new datasets were created or introduced.

Key Insights and Ideas

  • Paradoxical Response: The most striking finding is that an increase in tumor size on MRI after neoadjuvant therapy is associated with a favorable pathological response in UPS. This directly contradicts the conventional understanding that tumor shrinkage indicates treatment success.
  • Limitations of RECIST: The study strongly suggests that RECIST criteria, which focus on size reduction, are inadequate for assessing treatment response in UPS due to this paradoxical size increase.
  • Treatment Modality Matters: The pattern of size change differed between radiotherapy alone and chemoradiotherapy. While responders in both groups showed size increases, the chemoradiation group showed less pronounced increases, and non-responders in this group tended to shrink, though these differences weren't statistically significant.
  • Recurrence Clues: Non-responders who experienced local recurrence showed a reduction in tumor dimensions, unlike non-responders without recurrence.

Key Results

  • Increased Size in Responders: Pathological responders (pR) showed a significantly greater increase in tumor size across all measured dimensions compared to non-responders (pNR). For the longest axial diameter (X), pR showed a mean increase of 30% (±35%) versus 14% (±31%) for pNR, a statistically significant difference (p=0.003).
  • Area Increase: The cross-sectional area on the axial plane showed a significant difference, with responders experiencing a larger average increase (77% ± 112%) compared to non-responders (36% ± 75%), a difference of 41% (p=0.006).
  • Recurrence and Size Reduction: Non-responders who experienced local recurrence showed a reduction in all tumor dimensions compared to non-responders without local recurrence.
  • Inter-reader Agreement: Strong agreement (Kendall's W > 0.85) was observed among the four readers for tumor size measurements, validating the reliability of the imaging analysis.

Limitations and Future Work

The authors acknowledge several limitations:

  • Small Sample Size: The study included only 39 patients, which limits statistical power, particularly for subgroup analyses. This means the findings should be considered exploratory.
  • Retrospective Design: The retrospective nature of the study introduces inherent biases.
  • Single Institution: Data was collected from a single institution, limiting generalizability.
  • MRI Timing Variability: The interval between post-treatment MRI and surgery varied, which could affect the accuracy of tumor size measurements.
  • Focus on Size: The study primarily focused on size changes, excluding other multiparametric MRI markers.
  • Lack of External Validation: The developed optimal cutoffs require external validation before clinical application.

Future Work: The authors emphasize the need for larger, multi-institutional studies to confirm these findings and draw more definitive conclusions. They also suggest incorporating advanced functional imaging techniques like diffusion-weighted MRI (DW-MRI) and dynamic contrast-enhanced MRI (DCE-MRI) to provide a more comprehensive assessment of treatment efficacy and prediction of local recurrence. Developing subtype-specific prognostic and predictive biomarkers is also highlighted as a crucial next step.

Summary updated

Abstract

<b>Background/Objectives</b>: To correlate size changes in undifferentiated pleomorphic sarcoma (UPS) on magnetic resonance imaging (MRI) after neoadjuvant chemoradiation therapy (nCRT) with pathological response, risk of local recurrence, and therapeutic regimens. <b>Methods</b>: This retrospective study analyzed clinical, pathological, and imaging data from 39 biopsy-proven UPS subjects. Four readers measured the tumor dimensions before and after nCRT, including two perpendicular axial diameters and the longest coronal/sagittal diameter. Three cross-sectional areas and bounding volume were also calculated. Responders (pR) were defined as having ≤10% viable cells and non-responders (pNR) as having more. Inter-reader agreement was evaluated using Kendall's concordance coefficient. Changes in tumor size were compared between pR and pNR using one-way ANOVA and Tukey's HSD test for multiple comparisons of means. <b>Results</b>: pR showed a greater increase in size across all measurements compared to pNR. For the longest axial diameter, the mean increase was 30% ± 35% for pR and 14% ± 31% for pNR, with a mean difference (pR-pNR) of 16% (95% CI: 6-27%, <i>p</i> = 0.003). In tumors treated with radiotherapy alone, pR exhibited larger size increases in all dimensions compared to pNR. In contrast, in the chemoradiation group, pR showed a slight increase, while pNR generally shrank, although these differences did not reach statistical significance. Notably, pNR with local recurrence exhibited a reduction in all tumor dimensions compared to pNR without local recurrence. <b>Conclusions</b>: This exploratory study suggests that tumor size changes may predict pathological response and local recurrence after nCRT in UPS; however, the small sample size limits the generalizability of these findings.
BibTeX
@inproceedings{goreish2025paradoxical,
  author = {Mariam H. Goreish and Nicolò Gennaro and Laetitia Perronne and Görkem Durak and Amir A. Borhani and Hatice Savas and Linda C. Kelahan and Ryan Avery and Kamal Subedi and Tugce Agirlar Trabzonlu and Ulaş Bağcı and Baris Turkbey and Spyridon Bakas and Sean Sachdev and Ronen Sumagin and Borislav A. Alexiev and Pedro Hermida de Viveiros and Seth M. Pollack and Yury Velichko},
  title = {{Paradoxical Response to Neoadjuvant Therapy in Undifferentiated Pleomorphic Sarcoma: Increased Tumor Size on MRI Associated with Favorable Pathology}},
  booktitle = {Cancers},
  year = {2025},
  doi = {10.3390/cancers17050830},
}
Figure