Serum TARC dynamics during anti-PD1-based first-line Hodgkin lymphoma treatment: An analysis from the GHSG NIVAHL trial

Immune checkpoint inhibition (ICI) with anti-programmed death protein 1 (PD-1) antibodies is highly active as monotherapy in patients with relapsed classic Hodgkin lymphoma (cHL).^{1, 2} Recent studies combining nivolumab (N) with chemotherapy in first-line cHL treatment have demonstrated impressive outcomes in both early- and advanced-stage disease.^{3, 4} These studies did, however, not incorporate early response adaptations to guide treatment, but exposed all patients to full intensity chemo- and/or radiotherapy. This raises the question whether a one-size-fits-all approach is appropriate to reach the goal of minimizing treatment toxicity while maximizing efficacy in cHL. It can be envisioned that patients with an early and deep response to ICI-based first-line therapy are candidates for treatment de-escalation strategies, reducing exposure to chemo- and radiotherapy and resulting morbidity and mortality.⁵

To this end, recent studies incorporating other highly effective treatment regimens such as BrECADD (brentuximab vedotin, etoposide, cyclophosphamide, doxorubicin, dacarbazine, and dexamethasone) have demonstrated that response-adapted treatment enables de-escalation in patients with a negative interim 18F-FDG positron emission tomography (PET).⁴ However, the use of FDG-PET for response assessment to guide treatment has important limitations: Nonspecific uptake of 18F-FDG may result in a high number of false positives, particularly for patients with small tumor lesions classified as Deauville score (DS) 4.^{4, 6} The uncertainty due to false positivity seems even more important in the context of ICI, due to systemic inflammatory effects. Indeed, immune flare observed in 12%–23% of patients in the pivotal studies on nivolumab and pembrolizumab in cHL^{7, 8} led to the introduction of ICI-specific response criteria.⁹

Thymus and Activation Regulating Chemokine (TARC, or CCL17) has emerged as a biomarker for cHL disease activity. TARC is produced in large quantities by the malignant Hodgkin and Reed–Sternberg (HRS) cells and can aid in diagnosis using immunohistochemistry.¹⁰ In about 90% of patients, TARC is elevated in serum, with levels up to 400 times higher than those in healthy controls, and correlates with quantified FDG-PET results, particularly total metabolic tumor volume (TMTV).¹¹ Notably, serum TARC (sTARC) levels can be elevated years before clinical diagnosis.¹² In patients treated with ABVD, eBEACOPP, or salvage chemotherapies, sTARC dynamics correspond with response, and early sTARC reduction correlates with favorable outcome despite PET positivity.^{11, 13-15} Next to PET response, sTARC might hence be incorporated in future interim response-adapted strategies to tailor treatment intensity to individual need. However, to our knowledge, no data exist on the applicability of sTARC and its kinetics in first-line anti-PD1-based treatment.

In the prospective German Hodgkin Study Group (GHSG) randomized Phase II NIVAHL trial, patients with early-stage unfavorable HL were randomly assigned between concomitant N-AVD (nivolumab, doxorubicin, vinblastine, dacarbazine) or sequential N and (N-)AVD, each followed by 30 Gy involved-site radiotherapy (IS-RT). Outcomes in both treatment arms were excellent, also supported by quantified FDG-PET response using TMTV.^{4, 16} The aim of the current study was to evaluate sTARC dynamics during concomitant and sequential N-AVD treatment in NIVAHL and compare sTARC response with FDG-PET-based response assessment.

Patients in the randomized Phase II GHSG NIVAHL trial (NCT03004833) were evaluated for early treatment response by PET2 after either two cycles of concomitant N-AVD (Arm A) or four cycles of N monotherapy (Arm B) and after completion of systemic therapy as previously described.⁴ The current study includes all 78 patients out of 109 with informed consent, available serum samples at baseline and ≥1 additional timepoint, collected after 1 week of treatment, at PET2, after end of systemic treatment (EOT), and/or post-consolidative 30 Gy IS-RT (Table S1). Patients were excluded from sTARC response assessment if baseline sTARC was below the previously established¹⁵ and predefined cutoff of 1000 pg/mL (7/78). sTARC levels were measured using a standardized ELISA (R&D Systems, USA, Human CCL17/TARC Quantikine ELISA Kit) being blinded to treatment arm and response. Findings of sTARC were correlated with FDG-PET results using the DS and TMTV, as was previously described¹⁶ and to clinical outcomes. Differences in median sTARC levels across all timepoints were tested with Kruskal–Wallis, and a chi-square test was used to evaluate the association between categorical variables, with P < 0.05 considered statistically significant. All data were analyzed by GraphPad Prism 9.0 software. This study was conducted in accordance with the principles of the Declaration of Helsinki.

At baseline, elevated sTARC levels were observed in 71/78 patients (91%). The median sTARC level in these was 17,255 pg/mL (range: 1489–339,073 pg/mL). Compared to patients with sTARC elevation, patients without elevation were more often EBV+ (57% vs. 13%; P = 0.013). Baseline sTARC levels correlated with baseline TMTV in line with previous studies (Figure S1).^{11, 15} sTARC levels sharply decreased already 1 week after start of treatment in both treatment arms (Figure 1, P = 0.001). By the time of PET-2, only 8 of 65 (12%) patients with available samples at the time of PET-2 remained with elevated sTARC levels. At the EOT and after the end of treatment (i.e., after 30 Gy IS-RT), only 1 out of 56 patients (2%) and none out of 57 patients (0%) had sTARC above the threshold, respectively. Median sTARC levels at all timepoints during and after treatment were significantly lower compared to baseline sTARC values. Once normalized, sTARC levels did not increase above the threshold in any of the patients. In both treatment arms, sTARC normalization after 1 week or at first restaging was highly correlated with end-of-treatment PET negativity (P < 0.001).

During concomitant N-AVD treatment in Arm A, 8/34 patients (24%) remained FDG-PET positive with DS4 at interim-restaging after 2× N-AVD. sTARC was available in six of these patients; all of these patients had sTARC levels below the threshold. At EOT and after 30 Gy IS-RT, eight and four patients remained PET positive, respectively. Remarkably, all patients with available samples (seven and four) had sTARC normalization at these timepoints. None of these patients ultimately progressed or relapsed during a median of 41 months of follow-up, potentially indicating false-positive interim and EOT PET results in these patients using the DS4 cutoff (Figure 2A,B). The favorable early response pattern was further supported by a sharp decrease in mean TMTV of 97% at PET2 in 25 patients with available PET scans suitable for TMTV quantification.

In Arm B, after four infusions of N monotherapy, 13/37 patients (35%) remained PET positive with a DS of 4. Six out of 35 patients with available samples (17%) had sTARC levels above the threshold, with a partial response in five patients and stable disease in one by PET. Quantified PET results indicated a decrease in mean TMTV of 87% in sTARC-positive patients versus 99% in sTARC-negative patients at PET2, further supporting the deep response achieved after N monotherapy, reflected by both TARC and TMTV. Notably, the few patients with a sTARC normalization after 1 week of N monotherapy all had a negative PET at first restaging. After completion of systemic treatment with 2× N-AVD and 2× AVD, three patients still had a positive PET scan. Normalization of sTARC levels was observed in the single patient with an available sample. In the other two patients, sTARC level had already decreased to normal at first restaging in one patient and was normalized after radiotherapy in the other patient. All patients achieved a complete remission, reflected by both negativity of sTARC and PET after radiotherapy (Figure 2C,D).

One of the key observations from this first-ever study on sTARC dynamics during ICI-based first-line HL treatment is the rapid normalization of sTARC levels, regardless of whether nivolumab was administered concomitantly or sequentially with standard AVD chemotherapy. However, a slight difference in pattern between both treatment groups was observed, with earlier sTARC and PET normalization in the concomitant treatment arm. Summarizing results at first interim restaging in both study arms, sTARC was below 1000 pg/mL in 12/18 patients (67%) who still had a positive PET2 and available sample. Our results hence suggest that combining sTARC with PET imaging can help identify a relevant proportion of PET2-positive patients with a favorable prognosis early after the start of ICI-based first-line treatment. Next, all four patients with positive PET by DS at EOT had normal sTARC levels and did not show PD or relapse,¹⁷ further indicating the applicability of sTARC to reduce the number of false-positive PET scans, better tailor individual treatment exposure, and guide consolidative RT.

In light of the favorable response demonstrated by early sTARC negativity and sharp decrease in TMTV, our results imply that a relevant proportion of patients might be suitable for sTARC and PET response-based de-escalation strategies of anti-PD1-based first-line treatment. In early-stage cHL, these strategies might include the omission or reduction of chemotherapy and/or omission of radiotherapy, as is currently explored guided by MTV in the ongoing GHSG INDIE trial.¹⁸ In advanced-stage cHL, an intuitive strategy would be to reduce the number of N-AVD cycles in patients with early deep remission indicated by sTARC in light of quantified PET response at an interim restaging timepoint.¹⁹ Conversely—and as observed in some NIVAHL patients initially receiving nivolumab monotherapy—sTARC persistence during anti-PD1-based first-line treatment could indicate the need for more intensive treatment. Since the NIVAHL study was not designed in a response-adapted manner, no definitive conclusions can be drawn to support this approach. However, our findings do support the design and prospective evaluation of response-adapted anti-PD1-based treatment strategies in cHL using both sTARC and FDG-PET for response assessment. A potential limitation on the use of sTARC in response assessment is that a minority of patients (~10%) are TARC negative at baseline. Here, FDG-PET or assessment of measurable residual disease through circulating tumor DNA should be considered.²⁰

In conclusion, we observed high applicability of sTARC as a biomarker for treatment response during anti-PD1-based first-line treatment in cHL. Incorporating sTARC in response assessment in cHL has the potential to identify false-positive FDG-PET results. It might serve as an easily accessible, fast, and relatively cheap tool to explore individualized de-escalation strategies to further reduce treatment exposure and related morbidity in cHL patients.

Wouter J. Plattel: Conceptualization; investigation; funding acquisition; writing—original draft; methodology; formal analysis; project administration; data curation; supervision; resources; visualization. Sophie Teesink: Investigation; writing—original draft; methodology; formal analysis; data curation; visualization. Lydia Visser: Investigation; writing—review and editing. Conrad-Amadeus Voltin: Investigation; writing—review and editing; formal analysis; data curation. Helen Kaul: Investigation; writing—review and editing; methodology; formal analysis; data curation. Hans A. Schlößer: Investigation; writing—review and editing; resources. Bart-Jan Kroesen: Investigation; writing—review and editing. Carsten Kobe: Investigation; writing—review and editing; supervision. Bastian von Tresckow: Investigation; writing—review and editing; supervision. Peter Borchmann: Supervision; writing—review and editing; investigation. Arjan Diepstra: Investigation; conceptualization; funding acquisition; writing—original draft; methodology; formal analysis; supervision; data curation; resources; project administration. Paul J. Bröckelmann: Conceptualization; investigation; writing—original draft; funding acquisition; methodology; project administration; data curation; supervision; resources.

W.J.P. has received honoraria or travel support from Jansen-Cilag (institution) and Takeda (institution). B.v.T. is an advisor or consultant for Allogene, Amgen, BMS/Celgene, Cerus, Gilead Kite, Incyte, IQVIA, Janssen-Cilag, Lilly, Merck Sharp & Dohme (MSD), Miltenyi, Novartis, Noscendo, Pentixapharm, Pfizer, Pierre Fabre, Qualworld, Regeneron, Roche, SOBI, and Takeda; has received honoraria from AbbVie, AstraZeneca, BMS/Celgene, Gilead Kite, Incyte, Janssen-Cilag, Lilly, MSD, Novartis, Roche, Serb and Takeda; reports research funding from Esteve (insitution), MSD (insitution), Novartis (insitution), and Takeda (insitution); reports travel support from AbbVie, AstraZeneca, Gilead Kite, Janssen-Cilag, Lilly, MSD, Pierre Fabre, Roche, Takeda, and Novartis; and is member of steering committees for Regeneron (insitution) and Takeda. P.J.B. is an advisor or consultant for Hexal, MSD, Need Inc., Stemline, and Takeda; holds stock options in Need Inc.; has received honoraria from AstraZeneca, BeiGene, Bristol-Myers Squibb/Celgene (BMS), Eli Lilly, MSD, Need Inc., Stemline, and Takeda; and reports research funding from BeiGene (institution), BMS (institution), MSD (institution), and Takeda (institution). The other authors declare no potential COI.

The NIVAHL trial was sponsored by the University of Cologne and financially supported by Bristol-Myers Squibb. We thank the Anita Veldman Foundation for financial support of this study. P.J.B. is supported by an Excellence Stipend from the Else Kröner-Fresenius Foundation (EKFS). Open Access funding enabled and organized by Projekt DEAL.