Abstract PD5-05: Clinical implication of tumor infiltrating lymphocytes (TILs) in the ETNA study

• Published in: Cancer research (Chicago, Ill.) Vol. 80; no. 4_Supplement; p. PD5-05
• Main Authors: Bianchini, Giampaolo, Smart, Chanel, Mansutti, Mauro, Anton, Antonio, Licata, Luca, Sassi, Isabella, Calvo, Lourdes, Bisagni, Giancarlo, Bermejo, Begona, Semiglazov, Vladimir, Thill, Marc, Chacon, Jose Ignacio, Chan, Arlene, Murillo, Serafin Morales, Alvarez, Isabel, Lahuerta, Ainhara, Zucchinelli, Patrizia, Doglioni, Claudio, Viale, Giuseppe, Valagussa, Pinuccia, Tusquet, Ignasi, Gianni, Luca
• Format: Journal Article
• Language: English
• Published: 15.02.2020
• ISSN: 0008-5472; 1538-7445
• DOI: 10.1158/1538-7445.SABCS19-PD5-05

Abstract Background. We investigated the association between TILs assessed by different metrics and clinical outcome (pathological complete response -pCR- and risk of recurrence) in the ETNA trial. Methods. In the ETNA study (NCT01822314) 695 patients with HER2-negative breast cancer (BC) were randomized to receive neoadjuvant paclitaxel or nab-paclitaxel followed by 4 cycles of an anthracycline regimen. In the ITT study population, the two treatments did not show significantly different rates of pCR nor different Event-Free Survival (EFS) (Gianni JAMA Oncol 2018, Gianni ASCO 2019). We measured TILs by optical evaluation on pre-treatment core biopsies using three different metrics: stromal and intratumoral TILs (sTILs, iTILs), and hotspot TILs (hTILs)ie. the field of highest TILs density, in order to capture some of the spatial heterogeneity of TILs reported as more informative in ER+ BCs (Heindl JNCI 2018). We then investigated the association of TILs with pCR and EFS separately in TN and in LuminalB-like (ER+ and/or PgR+, Ki67≥14%) groups, as well as possible treatment-biomarker interactions. Results. TILs could be assessed in 589 pts (85%). The three TILs metrics were significantly correlated (p<10E-10) (sTILs vs iTILs, cor 0.59; sTILs vs hTILs, cor 0.83; iTILs vs hTILs, cor: 0.57). The correlation between hTILs and sTILs was lower for LumB-like tumors and lower TILs levels. iTILs were less informative than the other two metrics.Higher sTILs and hTILs as continuous variable (10% increase) were significantly associated with likelihood of pCR in both tumor subtypes (all p<0.001). In TN tumors sTILs and hTILs provided similar and not independent predictive value (OR 1.23 [1.10-1.37], p=0.002 and OR 1.34 [1.14-1.58], p=0.0005). In LumB-like tumors hTILs were slightly more predictive of pCR than sTILs. A 10% increase in hTILs was associated with a significant increase of pCR rate (OR 1.43 [1.28-1.61], p=1E-09). In addition, hTILs were more informative than sTILs (Likelihood Ratio Test p=0.01 for the addition of hTILs to sTILs in the model). In the TN BC group, higher expression sTILs and hTILs was associated with lower risk of recurrence, with HR 0.90 [0.81-0.99], p=0.041 and HR 0.84 [0.71-0.99], p=0.044, respectively, but in LumB-like BC they were not associated with EFS.We also evaluated the interaction between TILs and treatment benefit as exploratory analysis. The test for interaction was negative (p=0.251) in TN BC, while in the LumB-like group the test for interaction for hTILs and sTILs as continuous variable was 0.024 and 0.10, respectively. The test for interaction for sTILs was also significant when the transformed log variable was used (p=0.043), indicating a possible non-linear effect. To better evaluate the quality of this interaction, we applied an arbitrary cut-off to hTILs to define two groups: low (≤ 20%, n=282) and high (>20%, n=126) TILs. In the low hTILs group, abraxane was significantly more effective than paclitaxel (HR 0.64 [0.26-0.89], p=0.02). In the high hTILs group, paclitaxel was associated with better long-term outcome (HR 3.79 [1.33-10.8], p=0.01). If the same cut-off was applied in TN BC, abraxane had a numerical trend for superiority compared to paclitaxel in the low hTILs tumors (HR 0.64 [0.34-1.20]). No differences were observed in the high TIL group. Conclusions. Pre-treatment assessment of TILs in the ETNA trial was predictive of the likelihood of pCR regardless of the tumor subtype, and was prognostic in the TN subgroup. The assessment of hotspot TILs seems to be more informative than average stromal TILs in LumB-like tumors. Finally, different levels of hTILs at baseline were associated with different benefit from abraxane compared to paclitaxel. This finding warrants independent confirmation, and the mechanism for the association is being investigated. Supported in part by an unrestricted grant from Celgene Sarl, Swizerland Citation Format: Giampaolo Bianchini, Chanel Smart, Mauro Mansutti, Antonio Anton, Luca Licata, Isabella Sassi, Lourdes Calvo, Giancarlo Bisagni, Begona Bermejo, Vladimir Semiglazov, Marc Thill, Jose Ignacio Chacon, Arlene Chan, Serafin Morales Murillo, Isabel Alvarez, Ainhara Lahuerta, Patrizia Zucchinelli, Claudio Doglioni, Giuseppe Viale, Pinuccia Valagussa, Ignasi Tusquet, Luca Gianni. Clinical implication of tumor infiltrating lymphocytes (TILs) in the ETNA study [abstract]. In: Proceedings of the 2019 San Antonio Breast Cancer Symposium; 2019 Dec 10-14; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2020;80(4 Suppl):Abstract nr PD5-05.