Induction Chemo Turns Cold Tumors Hot in Metastatic TNBC

Marlene Kok, MD, PhD

Induction chemotherapy appears to make metastatic triple-negative breast cancer (TNBC) tumors more vulnerable to PD-1 blockade, according to results presented at the European Society for Medical Oncology’s 2018 Molecular Analysis for Personalised Therapy (MAP) Congress. Marlene Kok, MD, PhD, said results from the phase II TONIC study showed that induction therapy with doxorubicin turned “cold” tumor cells “hot,” resulting in an objective response rate (ORR) of 20% following treatment with nivolumab (Opdivo).

At a median follow-up of 15.2 months, there were 2 (3%) complete responses and 11 (17%) partial responses. The median duration of response was 9 months (95% CI, 2.7—not available).

“If you compare this with other trials…they report a response rate of about 5%,” said Kok, a study coauthor and a medical oncologist with the Netherlands Cancer Institute in Amsterdam. “This suggests that the induction treatments are bringing something.”

Even as investigators and physicians extend life for patients with breast cancer, the prognosis for those with TNBC remains poor, with a historical median overall survival (OS) of 8 to 13 months for metastatic disease.

Results from studies investigating PD-1 blockade in TNBC showed ORRs of 18.5% and 23.1% in studies of pembrolizumab (Keytruda) that selected patients based on positive PD-L1 status, findings that Kok called “promising in a disease with a poor prognosis.” In 2 studies that accepted patients regardless of PD-L1 status, the ORR dropped to 5.2% in a phase I study of avelumab (Bavencio) and 4.7% for a phase II study of pembrolizumab.

These results have presented 2 questions for those in the field of TNBC research: (1) How should anti—PD-L1 agents be combined with standard therapies? (2) How can the efficacy of anti–PD-L1 agents be improved? The TONIC trial set out to address those questions.

TNBC, Kok said, is immunologically a cold tumor. It is characterized by the presence of few T cells, most of which are nonclonal; an immunosuppressive tumor microenvironment; and a low number of antigens. A hot, or inflamed, tumor, by contrast, features many T cells and clonal T cells and no immunosuppression. Hot tumors are more sensitive to checkpoint inhibition therapy, so investigators are searching for ways to turn cold tumors hot.

Investigators in TONIC hypothesized that short-term induction with chemotherapy or radiation could modulate the immune response, leading to increased anti—PD-L1 activity. Kok said radiation activates the STING pathway, increases type I interferon interference, and directly stimulates T-cell priming, whereas chemotherapy with either cyclophosphamide, cisplatin, or doxorubicin could deplete regulatory T cells, increase type I interferon interference, and upregulate major histocompatibility complex class I molecules.

TONIC is a Simon’s 2-stage, adaptive, noncomparative phase II study analyzing 5 cohorts of 10 patients with paired biopsies. Patients were randomly assigned to induction therapy with 3 x 8 Gy radiation, 50 mg of daily cyclophosphamide for 2 weeks, 40 mg/m2 of twice-daily cisplatin, 15 mg of twice-daily doxorubicin, or a control group that received no induction therapy. All patients then received nivolumab for 1 year or until intolerable toxicity or disease progression.

Kok said a cohort stayed open until investigators collected 10 “good-quality” paired biopsies. “Otherwise, if you have 10 patients but only good-quality biopsies for 6, then the translation of research is really limited.”

The trial allowed early discontinuation of a cohort if ≤30% of patients achieved progression-free survival (PFS) of less than 12 weeks. Treatment cohorts showing strong clinical and translational outcomes would be expanded in the second stage of the study.

Primary clinical endpoints included PFS, OS, ORR, and clinical benefit rate. Translational endpoints included gene signature associated with PD-L1 response and number and quality of T cells.

Investigators took biopsies at baseline, after induction, and during nivolumab treatment. At the MAP Congress, Kok presented stage I final response data and first translational data.

Seventy patients took part in the study, with 68 included in the safety results and 67 included in the efficacy results. The median age was 51 years (range, 29-70). Twelve percent of patients had germlineBRCA1/2mutations, and 88% had wild-type mutations.

Most patients (49%) had received 1 prior treatment for metastatic disease, 29% had received 2 to 3 treatments, and 23% were treatment naïve. Ninety-one percent had received taxanes, 86% had received anthracyclines, 60% had received platinum, and 49% had received capecitabine.

By immunohistochemistry staining, 28% of patients were PD-L1—positive on <5% of immune cells, while 72% were positive on &ge;5% of cells.

The median ORR was 20% overall and highest in the doxorubicin arm, at 35%. The median ORR was 23% in the cisplatin arm, 17% in the control arm, and 8% in the radiation and cyclophosphamide arms.

&ldquo;The numbers are still small, but this suggests that the [response] might be different depending on the type of chemotherapy you use,&rdquo; she said. &ldquo;It is important to emphasize that we are not looking at response to cisplatin or doxorubicin because the dose is very low and all patients had [received] their anthracyclines before.&rdquo;

She added that most responses were observed during nivolumab treatment, suggesting that immune induction therapy might modulate the tumor microenvironment.

On T-cell receptor sequencing of biopsies performed during nivolumab treatment, the number of T cells was much higher in tumors that responded to anti—PD-L1 treatment. Clonality was also higher in those tumors.

Kok said that tumors treated with cisplatin or doxorubicin during induction had increased numbers of T cells and greater clonality—those cells were growing hotter. In contrast, tumors treated with radiation or cyclophosphamide, or those in the control group that did not receive induction therapy, did not experience a significant increase in T cells or clonality.

The data &ldquo;suggest that the cisplatin and doxorubicin might modulate the tumor microenvironment,&rdquo; she said.

Investigators next analyzed gene expression data for inflammation status using the NanoString IO 360 platform. Looking at the nivolumab biopsies, investigators reviewed 770 immune-related gene expressions and signatures. The analysis included 4 responders and 25 nonresponders.

By cohort, investigators found gene expression declining in the control and radiation arms in an analysis of biopsies taken after induction. &ldquo;Which is what we experience in the clinic,&rdquo; Kok said. &ldquo;The more disease progression, the less likely the patient [will] respond [to] anti—PD-1.&rdquo;

However, those hot gene signatures were upregulated in the chemotherapy groups, especially in the cisplatin and doxorubicin arms. The effect was even more pronounced in biopsies taken during nivolumab therapy. &ldquo;The tumors that have been exposed to cisplatin or doxorubicin are getting a hotter tumor microenvironment,&rdquo; Kok added.

&ldquo;We see that translational data go hand in hand with the clinical data, and both data suggest that if you use doxorubicin or cisplatin, you might increase the likelihood of response to anti—PD-1,&rdquo; she said. &ldquo;We&rsquo;ve seen upregulation of gene signatures that might be relevant and also increases in T cells and the clonality of the T cells.&rdquo;

The median OS at 1 year was 85% among responders, 50% among those who had stable disease at >24 weeks, and 15% for nonresponders.

Kok said that investigators are expanding the doxorubicin cohort in the second stage. Twenty patients have already been recruited.

Reference:

Kok M. Switching a cold into hot tumour: the TONIC trial. Presented at: MAP 2018 - Molecular Analysis for Personalised Therapy; September 14-15, 2018; Paris, France.