Joyce A. O’Shaughnessy, MD, led a group of physicians at a recent Targeted Oncology live case-based peer perspective event in discussion on emerging data that could affect treatment options for patients with triple-negative breast cancer in the future.
Joyce A. O Shaughnessy, MD
Joyce A. O’Shaughnessy, MD, led a group of physicians at a recentTargeted Oncologylive case-based peer perspective event in discussion on emerging data that could affect treatment options for patients with triple-negative breast cancer (TNBC) in the future. O’Shaughnessy, chair of the Breast Cancer Prevention Research, and co-chair of breast cancer research, Baylor Charles A. Sammons Cancer Center, The US Oncology Network, explained these emerging treatment options based on a case scenario of a patient with metastatic TNBC.
A postmenopausal 55-year-old African American woman was first given a diagnosis of breast cancer 2 years ago after discovering a lump in her right breast. She was active with 2 teenaged children.Imaging findings were suspicious for multifocal disease (largest, 2.5 cm) in the right breast and right axillary lymph node involvement. Pathology findings from ultrasound-guided core needle biopsy revealed a histologic grade 3 invasive ductal carcinoma. Her hormone receptor status was estrogen receptor negative, progesterone receptor negative, HER2 negative (immunohistochemistry, 0).Fine needle aspiration of a lymph node showed breast cancer. Her BRCA1/2 test results were negative.The patient underwent a mastectomy followed by immediate reconstruction; 2 of 20 nodes were positive. Surgical staging was T2pN1M0.Following surgery, she received adjuvant doxorubicin/cyclophosphamide followed by paclitaxel (12 weeks); she managed to complete chemo-therapy despite significant diarrhea and chemotherapy-induced nausea and vomiting.In routine follow-up at 18 months, the patient reported having worsening cough and abdominal pain. Imaging revealed 3 lesions in her right lung (<2 cm) and several liver lesions. A biopsy confirmed metastatic TNBC.
Targeted Oncology™: Would you wait for molecular testing results to come back before starting first-line therapy?
O’SHAUGHNESSY:I think it’s probably better to wait, only because the issue is, if you give some chemotherapy first, are you going to change the tumor microenvironment?
There was a preoperative study called the GeparNuevo trial, and it was in TNBC.1And they gave chemotherapy with durvalumab (Imfinzi), the antiPD-L1 inhibitor, versus just [nab-paclitaxel (Abraxane)] chemotherapy. But there was a third [“window-phase”] arm as well. They gave a 2-week run-in with durvalumab first. And then they came in, they kept durvalumab going and they added the chemotherapy. The pathologic complete response [pCR] difference was minor with the chemotherapy plus or minus durvalumab. But when you gave the durvalumab for 2 weeks, priming the immune system, it was a big difference, like a 20% difference [61.0% vs 41.4%]. That was weird.
So now they’re redoing that. So now AstraZeneca is going to try to do a run-in with antiPD-L1 first with durvalumab, and then come in afterwards with the chemotherapy. So I think I probably wouldn’t start without knowing [the test results] because [these patients] don’t benefit if they’re PD-L1 negative. I don’t want to give them [adverse] effects [given] that they might not benefit. So I would personally wait for it.
What are the therapeutic options at this point for a patient such as this?
Heretofore we haven’t had anything first-line that would be preferred over something else, so we didn’t have any survival advantage. So take your pick [in the] first line, right? I mean, the National Comprehensive Cancer Network [NCCN] says sequential single agents, you know, because you might as well reduce toxicity. Unless somebody is very, very ill, we’ve got to give them a doublet. But we didn’t have a survival advan­tage. But now we’ve got the atezolizumab [Tecentriq] and nab-paclitaxel [recommended by the NCCN] for the PD-L1 positive group.
There [are] lots of clinical trials going on. IMpassion131 [NCT03125902] might be [presented] at the San Antonio [Breast Cancer Symposium]. It’s paclitaxel plus/minus the atezolizumab in both intent-to-treat [ITT] and the PD-L1positive groups. We’re waitingwe don’t know, this might be [presented] at San Antonio; they’re awaiting [more progression] events. The KEYNOTE-355 trial [NCT02819518], that’s in first-line TNBC. Doctor’s choice is paclitaxel, nab-paclitaxel, or gemcitabine/ carboplatin, plus/minus pembrolizumab. We’re waiting on [the results of] that trial.
The IMpassion130 trial [was published in the] theNew England Journal of Medicine;2we’ve all seen this. [In the] ITT group, they did the PD-L1 testing and then they stratified patients. And we didn’t know whether [the patients] were PD-L1 positive or not until they stratified them…based on the PD-L1 status. [They also stratified patients in the trial based on] liver metastases (yes/ no) and prior taxane (yes/no). Patients got nab-paclitaxel weekly, plus or minus the every-2-weeks atezolizumab. The coprimary end points [were] progression-free survival [PFS] and overall survival [OS] in both the ITT and the PD-L1positive populations. So it’s really like 4 composite end points.
Forty-one percent of the breast cancers were PD-L1 positive.
In the ITT population, [the trial reached] significance. It was only a 1.7-month improvement in PFS in favor of the atezoli­zumab [7.2 vs 5.5 months; HR, 0.80; 95% CI, 0.69-0.92;P= .002]. And the statistical design said, “Thou shalt not look at PFS in a PD-L1positive [group] unless the PFS is positive in your ITT [group].”
Well it was positive, underwhelming but statistically signifi­cant. So they looked at the PD-L1positive, and lo and behold you get now 2½-months’ improvement [in PFS; 7.5 vs 5.0 months]. But the HR is better: it’s 0.62, so it’s betterit’s signif­icant [P<.001]. So there’s a modest improvement, but the disappointment [was that there was] no tail on that [survival] curve like you get in melanomalike you get in nonsmall cell lung cancer—[which] bummed us out. We wanted a tail on that curve. Disappointing.
So then, they looked at OS in the second interim analysis on survival, and you see there’s a 2½-month difference [21.0 vs 18.7 months; HR, 0.86], so it’s a modest improvement in survival but it’s not statistically significant3(P= .077).
So the statistical plan said, “Thou shalt not look at survival in the PD-L1positive [group] unless this is positive.” But because the PFS was positive only in the PD-L1 [group], they looked anyway. And that got the Europeans mad. They do not like it when you don’t follow the protocol. They think that’s the wrong thing to do.
The survival in the PD-L1positive population [shows] a 7-month improvement in OS [25.0 vs 18.0 months; HR, 0.71], and they don’t have aPvalue because it’s kind of considered exploratory. So that’s OS but it’s only in the PD-L1positive popu­lation. Maybe there’s going to be a little tail on that curvewe’re going to have to wait and see—we need longer follow-up to see the tail on the survival curve.
In terms of immune-related issues, the bottom line was that in the metastatic setting, you don’t see a high percentage of life-threatening colitis, myocarditis, encephalitis, pneumonitis, or hepatitis. You do see hypothyroidism and hyperthyroidism, but you see only a low level. We still have to be vigilant with 1% or 2%, but we don’t see a high level; it wasn’t a huge problem in the metastatic setting.
Would you recommend selecting treatment based on PD-L1 expression? How would you test for PD-L1 positivity?
This is new data from the European Society for Medical Oncoloy 2019 Congress. The 22C3 [PharmDx immunohistochemistry assay] is what’s used to choose patients for PD-L1 positivity for pembrolizumab [Keytruda]. And it turns out that 80% to 81% of TNBCs are positive by 22C3; 41% are positive by the [VENTANA PD-L1 (SP142) assay].4If you’re positive for SP142, you’re posi­tive for 22C3. But if you’re positive for 22C3, you’re not neces­sarily positive for SP142, because it’s only half of those who are going to end up being positive for both.
When this first came out, the labs didn’t have SP142, they were using 22C3. That was your standard of care. What happens if you go back to the IMpassion130 data, you get the block, and you look for the 22C3? There’s 80% of the TNBCs now are PD-L1 posi­tive. And then you take them and half of them got atezolizumab and half did not, and now you get a 2.4-month improvement in OS [21.6 vs 19.2 months] and a 2.1-month improvement in PFS [7.5 vs 5.4 months]. And with the SP142 assay, you get a 9.4-month improvement in OS [27.3 vs 17.9 months].
The SP142 is enriching for a much more definitive PD-L1 positive population. If we choose our patients with TNBC based on 22C3, this is what they’re going to get. We’re going to get lousy results; you know what I mean? So we don’t want to do that. So this was definitive, that we’ve got to use the SP142, or to get a 9-month improvement in survival, we’ve got to use SP142.
The pathologists know to do [PD-L1 testing on] breast cancers with SP142Caris Life Sciences, Inc., does SP142; Medfusion does it. Those are for sure. I haven’t used ProPath for the SP142. But you know what, this is better to write on the requisition and make sure it’s included.
If you look at primary breast cancers versus metastatic from the IMpassion130 trial, a higher percentage of the primary deposit for PD-L1 versus metastatic. The metastatic is like mid-40% versus mid-30%. So the metastatic lesions are not as likely to be PD-L1 positive, the primaries are more likely to be positive. That’s just an FYI.
What other newer treatment options are available for patients with TNBC and based on what data?
You’ll be interested in this study, the KEYNOTE-119 trial.5This was for patients with TNBC who had had 1 or 2 prior systemic therapies for metastatic disease, and they had to have had an anthracycline and taxane at some point. And then they did the PD-L1 [testing]. They didn’t choose the patients based on PD-L1 but they just did [the testing] first and then stratified them based on PD-L1. And they were randomized to single-agent pembrolizumab every 3 weeks or single-agent chemotherapy a physician’s choice. And the primary end point was OS.
[In the] ITT population there’s no benefit [in OS, 9.9 months with pembrolizumab vs 10.8 months with chemotherapy].
[They looked at PD-L1] with 22C3 [using] the composite posi­tive score, the CPS. What they do is they look for the percentage of positive tumor cells and the percentage of immune cellsand that means macrophages or lymphocytes—all over 100 cells. So the denominator is the number of cells and it’s a percentage of positive. [The trial looked at CPS groups of at least] 1%, 10%, and 20% [and each group showed an OS benefit with pembrolizumab]. That’s very interesting, isn’t it?
This will not get FDA approval because it missed its primary end point, plus the fact we are going to end up giving the chemotherapy with the checkpoint inhibitors. But this is very interesting. So, the more positive [a patient is for PD-L1], the better they’re going to do.
[With] chemotherapy, there’s no difference [in OS by CPS grouping]. The chemotherapy doesn’t care what the PD-L1 is. Of course, it’s not DNA-damaging therapy either. Platinum might care what the PD-L1 is, but those other chemotherapies don’t care, they’re not killing by immunologic cell death; you’re not inciting the immune system.
The KEYNOTE-522 trial is a really important trial [looking at neoadjuvant pembrolizumab with chemotherapy or chemother­apy alone followed by adjuvant pembrolizumab or placebo], and this is for newly diagnosed [patients with] stage II or III TNBC.6And, again, they were stratified for PD-L1 but not selected for PD-L1. And here the chemotherapy was carbopla­tin AUC [area under the concentration time curve] of 1.5, so a little light on the carboplatin, no breaks, versus AUC of 5 with weekly paclitaxel, followed by doxorubicin and cyclophospha­mide every 3 weeks, with pembrolizumab every 3 weeks versus placebo. Co-primary end points, pCR, and invasive eventfree survival [EFS].
It was a large trial; it had 1174 patients [included in a] 2:1 randomization.
Now, half and half, about 57%, got weekly low-dose carbopla­tin and 40% got heavy carboplatin.
Because it used a 22C3 antibody, 83% were PD-L1 posi­tive. Node involvement in about half. Clinically node positive or node negative.
The primary end point of improvement in pCR [in patients with] no disease in the breast or lymph node [showed a] 13.6 [percentage point] improvement in pCR. It’s around 14 [percent­age point improvement] no matter how you slice it, regardless of PD-L1 status.
The PD-L1positive patients had a higher pCR rate regardless [of whether] they got chemotherapy or pembrolizumab, and those who were PD-L1 negative didn’t do as well. [Patients just receiv­ing the] 4 chemotherapy drugs, they only got a 30% pCR rate, but it went it up 18 percent [points in the patients treated with added pembrolizumab] in the PD-L1–negative populationthat’s only 20%. And [there was] a 14 percentage point difference in the PD-L1–positive population [between chemotherapy with or without pembrolizumab].
Now looking carefully at nodal status at the pCR difference. So in node-positive disease, 20.6 percent [point difference with pembrolizumab] absolutely proven in pCR. Now we would all want that for our node-positive patients, 20 percent point more pCRs.
In the clinically node negative population, it was only 6.3 percent points difference in pCR because of the immune effects.
Look at the carboplatin. For the weak carboplatin, the AUC of 1.5, the pCR rate in the placebo was 48%. With the heavier carboplatin AUC of 5 every 3 weeks, the pCR rate was 8 percent points more, 56%. The problem is that’s going to be overlap­ping confidence intervals. But 48% versus 56%. But the delta, in terms of what the addition of the checkpoint inhibitor did if you added the checkpoint inhibitor to weekly carboplatin, it kind of made up for the weak carboplatin because it gave it an 18 percentage point improvement in pCR, whereas if you gave it the heavy carboplatin, it only had a 7.7 percent point improve­ment in pCR. Now these are subset analyses with overlapping confidence intervals.
It looks like, if this approved, it would be okay to give carbo­platin weekly because the pembrolizumab makes up for it. I personally would give it AUC 2, and then reduce the dose if you have to down to 1.5, but give a little extra.
The EFS is only with 15.5-months median follow-up [91.3% with pembrolizumab vs 85.3% with chemotherapy at 18 months]. Only 9% of the events have occurred of the planned events, which are recurrences. Very early data, but it’s impressive to have this 0.63 HR.
What is the safety profile like here?
You get some hypothyroidism, 15%. You get hyperthyroidism, 5%. These are going to be permanent. It looks like the thyroid [effect] is permanent. You get some severe skin reactions, 5%. Adrenal insufficiency, which is going to be permanent, is 2.7%, but not all that severe. Pneumonitis, which is reversible, 1.9%; colitis is reversible, 1.8%; hypophysitis, which is permanent, 1.8%. Thyroiditis and then hepatitis [are both] reversible. So that’s the toxicity, that’s the price, okay.
Do you think this treatment regimen will be made available to patients in the near future?
So here’s the summary: 13.6 percentage point improvement in pCR rate; a HR on EFS, with a very early follow-up, of 0.63. Immune, permanent life-long immune toxicity.
This was regardless of PD-L1 status, and this included node negative and node-positive patients. [Patients] could be stage II, could be T2N0 and go on this trial; we know T2N0 [patients] do well if it’s chemotherapy.
And so the dilemma is, what is the FDA going to do with this? Are they going to grant accelerated approval based on this short EFS follow-up, and then just give it a full whole approval if it turns [positively] out for EFS? And if they did, would the NCCN say the FDA doesn’t cherry-pick with regard to its subsets? But the NCCN could say, just use it for node positive, like they did [with the] AFFINITY trial, [where] the NCCN kind of says, use the pertuzumab [Perjeta] for node positive in the adjuvant setting.
[I think] the FDA probably won’t approve it for node [positive alone], because they don’t like subset analysis. If you don’t hit that primary end point, you’re not getting approved.
But they had survival data in the CLEOPATRA and in the IMpassion130 [trials where it was] only in the PD-L1positive group, and now we’re saying, approve it for everybody. Without PD-L1 there’s no selection fact that we can’t enrich.
I’m in the camp of I hope they give us an accelerated approval so we can pick and choose and give it to our higher-risked patients while we’re waiting. I’d rather save a few lives if it turns out to be positive than to incur the downside. If it turns out to be negative, you can cause some endocrine damage to people. I’d rather save a few lives. If I’m going to err on something, I’d rather err on the side of not having lost somebody because I should have given [something], but everybody is different. Anyway, they’ll decide for us.