Analyzing ADC Options Across Metastatic TNBC

In a Satellite Session focused on antibody drug conjugates (ADCs) in metastatic triple negative breast cancer (TNBC), clinicians from Memorial Sloan Kettering (MSK) Cancer Center, met to discuss the efficacy, safety, and real-world implementation of these treatments. The team dove into results from the phase 3 ASCENT-03 trial (NCT05382299), phase 3 ASCENT-04 trial (NCT05382286), phase 3 TROPION-Breast02 trial (NCT05374512), and phase 3 DESTINY-Breast04 trial (NCT03734029).^1-4 Within each of the trials, the panelists looked at the agents of sacituzumab govitean-hziy (Trodelvy), datopotamab deruxtecan-dlnk (dato-DXd; Datroway), and fam-trastuzumab deruxtecan-nxki (T-DXd; Enhertu).

For this portion of the discussion, Tiffany Traina, MD, FASCO, vice chair of Outpatient Operations, Department of Medicine, and section head of the Triple Negative Breast Cancer Clinical Research Program, led the charge. She was joined by Monica Fornier, MD, associate attending physician; Mila Gosky, MD, associate attending physician; Gabriella D'Andrea-Carlino, MD, a medical oncologist; Devika Gajria, MD, Medical Site Director of MSK Bergen; Diana E. Lake, MD, attending physician; and Amy Xu, MD, director of Metastatic Research Program in the Department of Radiation Oncology.

ASCENT-03 Data

Traina: I’m going to start with ASCENT-03, which is the sacituzumab trial for first-line metastatic TNBC. A few things to call out that are differentiators. This study enrolled patients who were not candidates for PD-L1 inhibitors. That could mean they had PD-L1 negative breast cancer. It could mean they had a contraindication to a checkpoint inhibitor. It could mean that they’re in a place globally where they didn’t have access to a checkpoint inhibitor. For whatever reason, they’re not considered candidates for pembrolizumab (Keytruda). The other key thing is their disease-free interval was 6 months or greater from adjuvant therapy. The randomization was 1:1 sacituzumab vs chemotherapy. The chemotherapy arm here included either taxane or gemcitabine/carboplatin as a doublet. An important design feature here is that if patients were randomly assigned to the control and progressed, they were permitted to cross over to sacituzumab.

The other key distinction in this trial is that the primary end point is only progression-free survival (PFS). Overall survival (OS) is not powered to show a difference. By nature of the design with this crossover, even if a difference were there and we had the power to see it, it might be diluted by the fact that patients could get sacituzumab in a later line setting where there’s an OS advantage, and that might dilute out any OS difference seen. About two-thirds of these patients are White. Two-thirds of the patients were treated outside of the US, Canada, and Europe. Most of these patients had a PD-L1 negative tumor. A total of 99% or greater were PD-L1 negative. The disease-free interval breakdown is an important one. A third of patients had de novo metastatic disease. Additionally, 20% of patients were progressing within 6 to 12 months of their adjuvant therapy. [Most patients] with visceral sites of disease and almost everybody had seen prior taxane. Very few had seen prior platinum-[based therapy].

PFS was significantly improved with sacituzumab over standard-of-care chemotherapy. The medium was PFS of 9.7 months compared with [6.9] months with chemotherapy, a 40% improvement in PFS and that met the primary end point. When you look across all the subgroups, essentially, they favor the ADC over chemotherapy. The OS is going to be difficult to conclude anything about. At this first analysis, the data are immature. It’s too soon to look at this. Also, remember there’s crossover. It’s not powered to show a difference. At this point, there’s no difference seen, and the medians haven’t been reached. PFS2 was used as a bit of a surrogate for long-term outcomes.

At a minimum, you’re seeing that the group that [received] sacituzumab were doing no worse than the group that [received] chemotherapy. It looks like they’re doing a bit better, having seen sacituzumab in the first-line setting. That’s encouraging for what you can conclude there. It’s important to note that in that crossover that was built in, about 80% of patients [who] were in the chemotherapy group who went on to subsequent second-line treatment received sacituzumab. There is a significant number of patients [who] went on to second-line sacituzumab. If you look at the proportion, a different way of looking at these data is what proportion of patients in the overall control arm have gone on to get sacituzumab and that’s about 50%. That considers the proportion of patients [who] might still be on treatment, on chemotherapy-derived benefits, so they haven’t yet had to make a decision –– or the 20% of patients [who] go on to second-line therapy but didn’t [receive] sacituzumab.

[Another] important efficacy end point is response rate. Somewhat surprisingly, the overall response rate with sacituzumab was identical to that of chemotherapy. No difference in complete response proportion, that’s 5% to 7%. No difference in partial response, that’s about 40% [between both arms]. Of those who did respond, they had a much more durable benefit if they were getting sacituzumab than chemotherapy. The median duration of response was about a year with sacituzumab and only 7 months with chemotherapy. What is your impression, of seeing the ASCENT-03 [data], how [would] you interpret OS. What are your takeaways?

Carlino: I’m impressed with the efficacy. The crossover design makes it hard to appreciate whether there’s going to be a survival benefit. This was a sick population. You’re seeing this meaningful PFS. It’s a good surrogate for OS.

Fornier: The PFS2 [data] offers additional context because we’ve seen in other trials, not for ADCs or for chemotherapy, that sometimes you would see a benefit with PFS1, but then it cancels out PFS2. Instead, the fact that the benefit was maintained for a patient who got sacituzumab considering also that over 80%, went on to get sacituzumab as a second line. It does offer additional context as showing that this drug is truly better in the first line as opposed to standard chemotherapy.

Gorsky: It’s impressive. I started using it as a first line already and I find [patients] do well with it.

ASCENT-04 Data

Traina: I want to jump into ASCENT-04, which is the sister study to ASCENT-03. This is looking at first line metastatic PD-L1 positive TNBC. Here prior checkpoint inhibitor was allowed, but again, because of the timing, it’s going to be very few patients [who] have seen a checkpoint inhibitor. This study has that same 6-month disease-free interval, just like ASCENT-03. It’s a 1:1 randomization of sacituzumab/pembrolizumab or chemotherapy/ pembrolizumab, just like the phase 3 KEYNOTE-355 (NCT02819518) regimen, taxane, or gemcitabine/carboplatin. Crossovers were built in and allowed the same way it was in ASCENT-03, and the primary end point is PFS alone, just like ASCENT-03.

Patient characteristics also look very much like ASCENT-03, although here 100% of patients have PD-L1 positive tumors. About a third have de novo metastatic disease. There’s a cap of around 20% or so that were recurring within 6 to 12 months. The other half of the patients are recurring beyond a year. Visceral disease was seen, only about 5% of patients having seen a prior checkpoint inhibitor. The taxane vs gemcitabine/carboplatin split was about 50/50.

The median PFS was 11.2 months [in the sacituzumab arm vs] chemotherapy/pembrolizumab was 7.8 months. There was a 35% improvement in PFS, with a statistically significant and clinically meaningful difference here. When you look at the preplanned subgroup analyses, almost all of them are favoring sacituzumab/pembrolizumab. You could see in the middle there you’ve got that teeny subset of patients who had prior exposure to a checkpoint inhibitor, but that’s only 20 patients. I don’t think we can conclude anything about checkpoint inhibitor after checkpoint inhibitor.

The OS data are immature at this analysis. You’ve got crossover baked in here, so it makes it challenging to see and it’s not a primary end point. It’s not powered to show a difference. It doesn’t appear that there’s any difference here. About 80% of patients in the control arm were able to go on to get sacituzumab upon progression. [R]esponse rates were a bit higher with sacituzumab/pembrolizumab, then chemotherapy/pembrolizumab at 60% response rate over 53%. We also see in these numbers the prognostic value of having a PD-L1 positive tumor, because all these numbers are better than in our PD-L1 negative [patients]. Response rates are a lot higher; duration of response is longer. This is a better prognostic group of patients than when you have a PD-L1 negative TNBC. What do you take away in the PD-L1 positive space? Are there any advantages, thoughts that you have around this combination? Is this your next go-to regimen now in the first line setting?

Gajria: It would be my first line regimen. Unfortunately, I haven’t had anyone. I’ve just had a lot of KEYNOTE[-355], I don’t think anyone’s progressed yet. In this situation, everyone would’ve seen pembrolizumab before, but of course if I had a de novo patient this would be the way to go. For me, this is like an easy bifurcation in the pathway. I understand if they’re PD-L1 positive and go this route if they are because you know you’re going to improve their outcomes.

Traina: What window would you use, that “magical” 1-year mark? Say if they recurred within a year of their last pembrolizumab [dose], you’re not going to bother rechallenging, but if it’s longer than a year, then we will try it again.

Fornier: It’s like what we used to do with chemotherapy. Let’s say within a year of the last taxol, then you would rechallenge with weekly taxol. If it’s been like 5 months or 6 months or something, then say, ‘Well, probably it didn’t work. It’s best to use something different’. I would follow that guideline in the absence of clear data in this setting.

Traina: I have to say, I’m surprised that we don’t have these data yet because we’ve had checkpoint inhibitors in metastatic disease for a sufficient amount of time. Trials that have been designed to look at checkpoint inhibitor after checkpoint inhibitor in metastatic disease. We haven’t seen those results, which is a little frustrating. We’re trying to extrapolate what we do [for] early stage to first-line metastatic. I wish we had these data to guide us.

Carlino: I don’t think we could possibly know what would happen. Someone progressing on pembrolizumab is incredibly discouraging, but somebody that’s progressing 4 months, 6 months, I don’t think we know what that means.

Lake: I agree. It’s such a poor prognosis. I probably would hold the pembrolizumab and do another ADC.

Xu: From a radiation context, usually I’m seeing them if they have brain metastases and I’m just curious how that influences which ADC you would start with or sequencing.

Traina: There’s some limited central nervous system [CNS] data for sacituzumab as well as for dato-DXd. For sacituzumab, if you looked at the ASCENT trial, the original phase 3 ASCENT trial (NCT02574455) allowed patients who had brain metastases. The median PFS for those patients with brain metastases was just under 3 months. It was like 2.8 months, and chemotherapy was 1.6 months… In TROPION-Breast02 that we’re about to talk about 10% of that patient population had brain metastases. Not only did they have stable treated brain metastases, but they could also have untreated asymptomatic brain metastases, almost like the HER2CLIMB [patients]. It’ll be interesting to see. It’ll be small numbers because it’s only 10%, but you’ll have some patients in there [who] had untreated CNS disease, and it would be great to see CNS response from dato-DXd. That would be encouraging.

TROPION-Breast02 and DESTINY-Breast04 Data

Speaking of TROPION-Breast02, why don’t we chat about these data? I’ll call out the differences here now for first-line metastatic TNBC PD-L1 negative or patients not a candidate for checkpoint inhibitor. That could mean again, toxicity, contraindication, PD-L1 negative, somewhere globally where they didn’t have access to the drug, or maybe the physician felt that they were such a rapid progressor on checkpoint inhibitor that they weren’t a candidate to continue checkpoint inhibitor. One of the most impressive pieces of this study is that there was no minimum disease-free interval. You have [patients] recurring on their adjuvant therapy or just within a couple of months of adjuvant therapy. Our worst players were allowed on the study, and you could have CNS disease. There was no crossover. Patients were [randomly assigned] to either dato-DXd as a single agent once every 3 weeks or investigator choice chemotherapy, which is a bit of a misnomer because the FDA required what you got as standard chemotherapy based on certain clinical criteria. If you had your patients [randomly assigned], if they had never seen a taxane, they needed to get taxane for control. If they had a long disease-free interval of more than a year, they needed to get a taxane. It was just the rapid progressors that had previously seen a taxane [who] were able to get something else, and that something else could have been single-agent carboplatin, single-agent eribulin [Halaven], or single-agent capecitabine.

Another important difference in this trial is the statistical design was a dual primary end point for PFS and OS. You may wonder why a crossover wasn’t there, but for this trial and this agent, dato-DXd, it certainly didn’t have a label indication, and it doesn’t have a survival benefit proven in late-line triple-negative breast cancer the way sacituzumab did. The only indication that came out while this trial was going on was the hormone receptor-positive indication. It almost feels like an unfair expectation to offer crossover when they didn’t have that same data to say there was a benefit to getting the drug in a later line. I will share with you the data about post-progression ADC of any sort that was given to the control arm.

These are the baseline characteristics. There’s more racial/ethnic diversity in this patient population. A total of 47% of patients identified as Asian, and 41% were White. The geographic region again is about two-thirds outside of Canada/US/Europe, and about 10% of patients had PD-L1 positive tumors. That’s higher than what was in ASCENT-03. For the disease-free interval, you have 15% of patients recurring on their adjuvant therapy or within 6 months of adjuvant therapy. While we thought ASCENT-03 were patients who were high-risk, these are even more enriched for those highly refractory patients. There’s a 20% cap for disease-free interval of under a year. Almost 80% of patients had visceral metastases and large burden of disease primarily. Only 5% of patients had seen prior checkpoint inhibitor.

The first primary end point, PFS by blinded independent review showed a significant, meaningful difference. The median PFS is 10.8 months (95% CI, 8.6-13.0). With chemotherapy, it’s 5.6 months (95% CI, 5.0-7.0). An absolute difference of 5 months in PFS and a 43% improvement in PFS. When you look at all the subsets that are pre-planned for PFS, they all favor dato-DXd over chemotherapy.

There’s about a 20% improvement in OS. The median OS is approaching 2 years, 23.7 months (95% CI, 19.8-25.6), and chemotherapy about a year and a half. That incremental absolute difference of 5 months between the 2 arms. When you look at the OS subgroup analysis, they’re all trending towards dato-DXd. The one that has raised some eyebrows in question is this group here, which is representing US, Canada, and Europe. At the presentation at European Society of Medical Oncology [ESMO] 2025, when Rebecca [Dent, MD, MSc], presented our data, there was an unplanned further subset analysis of just the US population. The reviewers asked us to do this unplanned analysis because, in the US, patients have access to post-progression ADC. The overall HR approximated that of the intention-to-treat population, favoring dato-DXd over chemotherapy. I don’t know how to explain the Canada/Europe piece and what’s pulling towards chemotherapy, but in the US population, the HR again was looking like a 20% OS advantage.

[Let’s look at] DESTINY-Breast04, about T-DXd and HER2-low in later line treatment. This is largely a hormone receptor-positive study. Patients were [randomly assigned] to T-DXd or standard of care chemotherapy. They needed to have had prior exposure to endocrine therapies and up to 2 prior lines of chemotherapy. The primary end point is PFS. Here, it’s looking at just the 60 patients [who] had hormone receptor negative- HER2-low breast cancer, with a significant benefit to T-DXd over chemotherapy. The median PFS was 6.3 months (95% CI, 4.2-8.5) vs 2.9 months (95% CI, 1.4-4.2). The median OS was 17.1 months (95% CI, 13.6-23.0) compared with 8.3 months (95% CI, 6.6-20.4). This is looking consistent with what the bigger trial population showed, but it is just a small subset of a bigger trial.

Lake: The options are reasonable, and my decision would be based on toxicity.

Traina: The other thing I didn’t mention in here, but we’ve talked about it offline and in our service meetings, is the schedule differences and the chair time difference.

Fornier: There’s the fact that the schedule is different, make it either 1 way or another. Do you think it’s 2 weeks on, 1 week off? You can manage better, for example, you can skip day 8 rather than giving every 3 weeks, or do you think every 3 weeks is better because patients have to come less frequently?

Traina: Just from a patient experience standpoint, it’s nicer to be able to come once every 3 weeks. Any deviation from the 2 [weeks]-on, 1 [week]-off is something that’s investigational. Even though we tend to do it over time and make up in every 2-week regimen or once every 3 with sacituzumab, it's not necessarily the best. There’s an ongoing trial testing sacituzumab govitecan every 2 weeks, but the dose is higher than the dose we use for day 1, day 8. That’s pharmacokinetic-driven trying to get to an easier schedule might give us higher toxin another way if we’re giving a higher dose.

References

1. Cortés JC, Bardia A, Punie K, et al. Primary results from ASCENT-03: a randomized phase III study of sacituzumab govitecan (SG) vs chemotherapy (chemo) in patients (pts) with previously untreated advanced triple-negative breast cancer (TNBC) who are unable to receive PD-(L)1 inhibitors (PD-[L]1i). Abstract presented at: European Society for Medical Oncology Congress 2025; October 17-21, 2025; Berlin, Germany. Abstract LBA20.
2. Tolaney SM, de Azambuja E, Kalinsky K, et al. Sacituzumab govitecan plus pembrolizumab for advanced triple-negative breast cancer. N Engl J Med. 2026;394(4):354-366. doi:10.1056/NEJMoa2508959
3. Dent R, Shao Z, Schmid P, et al. Datopotamab deruxtecan in patients with untreated, advanced triple-negative breast cancer (TROPION-Breast02): a randomised, open-label, international, phase III trial. Ann Oncol. Published online April 3, 2026. doi:10.1016/j.annonc.2026.03.008
4. Modi S, Jacot W, Iwata H, et al. Trastuzumab deruxtecan in HER2-low metastatic breast cancer: long-term survival analysis of the randomized, phase 3 DESTINY-Breast04 trial. Nat Med. 2025;31(12):4205-4213. doi:10.1038/s41591-025-03981-4