📄 medicine

Mid-treatment radiologic response as an early predictor of pathologic response in TNBC in the era of pembrolizumab

In this retrospective study of 517 triple-negative breast cancer patients, mid-treatment radiologic complete response emerged as a stronger predictor of pathologic complete response than post-treatment imaging, with pembrolizumab demonstrating the highest likelihood of inducing early radiologic response.

Original authors: Soong June Bae, Na Lae Eun, Jee Hung Kim, Yoonwon Kook, Ji Hyun Youk, Jeong-Ah Kim, Eun Ju Son, Joon Jeong, Yoon Jin Cha, Sung Gwe Ahn

Published 2026-07-01

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Soong June Bae, Na Lae Eun, Jee Hung Kim, Yoonwon Kook, Ji Hyun Youk, Jeong-Ah Kim, Eun Ju Son, Joon Jeong, Yoon Jin Cha, Sung Gwe Ahn

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

The Big Picture: Checking the Scoreboard Early

Imagine Triple-Negative Breast Cancer (TNBC) as a very aggressive, fast-moving storm. Doctors treat this storm with "Neoadjuvant Chemotherapy" (NAC), which is like sending in a massive fleet of weather-control ships to calm the storm before the patient needs surgery.

The ultimate goal is Pathologic Complete Response (pCR). In our analogy, this means the storm has completely vanished by the time the ships arrive at the harbor. If the storm is gone, the patient has a much better chance of a long, healthy life.

For a long time, doctors had to wait until the very end of the treatment (just before surgery) to see if the storm had cleared. This new study asks a crucial question: Can we check the "weather radar" halfway through the treatment to see if the storm is already gone?

The Study Setup: Three Different Storm-Fighting Fleets

The researchers looked at 517 patients treated at Gangnam Severance Hospital over a long period. They compared three different types of "fleets" (treatment regimens) used to fight the cancer:

The Standard Fleet: Traditional chemotherapy drugs (no special additions).
The Heavy-Duty Fleet: Traditional drugs plus Carboplatin (a stronger, more toxic drug).
The High-Tech Fleet: Traditional drugs plus Carboplatin and Pembrolizumab (a modern immunotherapy drug that helps the body's own immune system fight the cancer).

The researchers used MRI scans (a super-powered camera) to take pictures of the tumors at three times:

Start: Before any treatment.
Mid-treatment: Halfway through the therapy.
End: Just before surgery.

The Key Findings

1. The "Mid-Treatment" Checkpoint is a Crystal Ball

The study found that if the MRI halfway through treatment showed no visible cancer left (called a "radiologic complete response" or rCR), it was a massive predictor that the cancer would be completely gone by the time of surgery.

The Analogy: Imagine a race where you check the runner's position at the halfway mark. If they are already at the finish line at the halfway point, they are almost guaranteed to win the race.
The Data: Patients who looked "clear" on the mid-treatment MRI were 26 times more likely to have a completely clear tumor at surgery compared to those who still looked like they had cancer. This was a stronger predictor than even the final MRI taken right before surgery.

2. The "High-Tech Fleet" (Pembrolizumab) Wins the Early Race

When comparing the three fleets, the one with Pembrolizumab (the immunotherapy) was the most effective at clearing the cancer early.

The Analogy: Think of the three fleets as different types of fire extinguishers. The standard one puts out some fire. The heavy-duty one puts out more. But the High-Tech one (Pembrolizumab) was the only one that put out the fire so fast that, halfway through the job, the smoke was already gone for nearly a quarter of the patients.
The Data:
- Pembrolizumab group: 24.3% looked clear halfway through.
- Carboplatin group: 12.8% looked clear.
- Standard group: 11.6% looked clear.

If a patient on the Pembrolizumab plan looked clear halfway through, there was a 97% chance they would have zero cancer left at surgery.

3. The "Late Bloomers"

The study also noted that some patients who didn't look clear halfway through eventually cleared up by the end. However, this happened less often (about 1 in 5 patients), and even then, they didn't always achieve a "perfect" clear-out at surgery. This suggests that waiting for a "late bloom" is riskier than catching an "early bloom."

What the Authors Are (and Are Not) Saying

The researchers are very careful about what they claim.

What they claim: Mid-treatment MRI is a powerful tool to predict if the cancer is gone. Pembrolizumab seems to make the cancer disappear faster than other drugs.
What they DO NOT claim: They do not say that doctors should stop treatment early just because the MRI looks clear.
- Why? The study was "retrospective," meaning they looked back at old records. In those old records, everyone finished the full treatment, even if they looked clear halfway through.
- The Caveat: The authors explicitly state that we don't know yet if it's safe to stop treatment early. They say these findings are "hypothesis-generating." This means the study is a strong suggestion that we should run new, forward-looking experiments to see if we can safely shorten treatment for people who respond this well.

The Bottom Line

This paper is like a pilot test for a new navigation system. It shows that checking the map halfway through the journey (Mid-treatment MRI) is incredibly accurate at predicting if you've reached your destination (No cancer at surgery). It also suggests that the newest vehicle (Pembrolizumab) gets you to that "clear" status faster than the older models.

However, the authors warn that just because the map says "destination reached" halfway through, we can't yet change the rules of the road to stop driving. We need new, planned experiments to prove that stopping early is safe. For now, the study provides a strong reason to design those future experiments.

Technical Summary: Mid-treatment Radiologic Response as an Early Predictor of Pathologic Response in TNBC in the Era of Pembrolizumab

Problem Statement
Triple-negative breast cancer (TNBC) is an aggressive subtype with limited treatment options, where neoadjuvant chemotherapy (NAC) is the standard of care for stage II–III disease. While achieving a pathologic complete response (pCR) correlates with favorable long-term outcomes, current pCR rates with conventional regimens remain modest (30–40%). The introduction of intensified regimens incorporating platinum agents (carboplatin) and immune checkpoint inhibitors (pembrolizumab), such as in the KEYNOTE-522 trial, has improved pCR rates but significantly increased toxicity and treatment burden. Consequently, there is a critical need for predictive biomarkers to guide treatment tailoring and avoid overtreatment. While breast magnetic resonance imaging (MRI) is used to monitor NAC response, its predictive value for pCR specifically in the modern era of intensified chemoimmunotherapy remains incompletely defined. This study addresses the gap in understanding the association between mid-treatment radiologic complete response (rCR) and subsequent pCR across different NAC regimens.

Methodology
This was a retrospective, single-center cohort study conducted at Gangnam Severance Hospital involving 517 patients with stage II–III TNBC treated between January 2008 and March 2025. The inclusion criteria required patients to receive at least 75% of planned NAC cycles and undergo breast MRI at three specific time points: baseline, mid-treatment (halfway through NAC), and post-NAC (within one month prior to surgery).

Patients were categorized into three treatment cohorts based on their NAC regimen:

Non-carboplatin: Anthracycline- and taxane-based combinations without carboplatin or pembrolizumab ( $n=225$ ).
Carboplatin: Regimens including paclitaxel and carboplatin, followed by anthracycline/cyclophosphamide ( $n=148$ ).
Pembrolizumab: The KEYNOTE-522 regimen combining the above chemotherapy backbone with pembrolizumab ( $n=144$ ).

Radiologic response was assessed via breast MRI. Radiologic complete response (rCR) was defined as the absence of measurable residual enhancement at the primary tumor site and no suspicious axillary enhancement, adapted from RECIST v1.1. Pathologic response was determined by surgical specimens, with pCR defined as the absence of residual invasive carcinoma and carcinoma in situ in both the breast and axillary lymph nodes (ypT0N0). Statistical analyses utilized binary logistic regression to evaluate associations between radiologic responses and pCR, adjusting for age, tumor grade, clinical stage, and nodal status.

Key Contributions and Results
The study analyzed 517 patients, of whom 225 (43.5%) achieved pCR. Key findings include:

Regimen-Specific rCR Rates: Mid-treatment rCR rates varied significantly by regimen. The pembrolizumab group exhibited the highest rate (24.3%), followed by carboplatin (12.8%) and non-carboplatin (11.6%). Multivariable analysis confirmed that pembrolizumab was independently associated with higher odds of mid-treatment rCR (adjusted OR 3.33; 95% CI 1.74–6.34; $p < 0.001$ ) compared to non-carboplatin regimens.
Predictive Value of Mid-treatment rCR: Mid-treatment rCR was a robust predictor of pCR across all regimens. Patients achieving mid-treatment rCR had a significantly higher likelihood of achieving pCR (adjusted OR 26.33; 95% CI 9.26–74.81; $p < 0.001$ ). This association was stronger than that of post-NAC rCR (adjusted OR 8.85; 95% CI 5.52–14.18).
High Positive Predictive Value: Among the 80 patients (15.5%) who achieved mid-treatment rCR, 73 (91.3%) subsequently achieved pCR. This rate was particularly high in the pembrolizumab group (97.1%).
Dynamics of Response: Most patients with mid-treatment rCR maintained this status through post-NAC imaging and achieved pCR. Conversely, among patients without mid-treatment rCR, approximately 18.6–24.8% demonstrated a "late response" (conversion to post-NAC rCR), but only about 54–70% of these late responders ultimately achieved pCR.

Significance and Claims
The authors posit that mid-treatment rCR serves as a strong early predictor of pCR in TNBC, regardless of the specific NAC regimen used. The study suggests that the addition of pembrolizumab is associated with a higher likelihood of achieving this early radiologic response.

However, the paper explicitly frames these findings as hypothesis-generating rather than definitive proof of causal effects or survival benefits. The authors emphasize several limitations:

The retrospective, single-center design introduces potential selection and information biases.
The three regimens were administered during different calendar periods, meaning comparisons are confounded by era, evolving patient selection, and changes in imaging technology.
The study did not analyze long-term survival outcomes.

Consequently, the authors conclude that while their data supports the rationale for prospective, response-adapted trials, it cannot yet confirm whether de-escalating NAC (e.g., omitting sequential chemotherapy) based on mid-treatment MRI response is safe or effective. The primary significance of the work lies in providing a rationale for future prospective studies to evaluate whether mid-treatment MRI can inform individualized treatment strategies and de-escalation in TNBC patients receiving modern chemoimmunotherapy.