Neoadjuvant Immune Checkpoint Blockade Enhances Local and Systemic Tumor Immunity in Head and Neck Cancer

Abstract

Purpose of review

Neoadjuvant (pre-surgical) immune checkpoint blockade (ICB) has shown promising clinical activity in head and neck cancer and other cancers, including FDA approvals for neoadjuvant approaches for triple-negative breast cancer and non-small cell lung cancer. Here we will review recent data from clinical trials in HNSCC, including mechanistic studies highlighting local and systemic effects on T cell-mediated immunity.

Recent findings

A series of clinical trials of neoadjuvant ICB have documented evidence of clinical activity, including clinical to pathologic downstaging and pathologic response in a subset of patients. Also, emerging data suggest improved survival outcomes for patients with tumors responsive to neoadjuvant ICB. In depth mechanistic studies have documented intra-tumoral expansion of CD8 T cell populations characterized by tissue residency and cytotoxicity programs. Treatment also leads to expansion of activated CD8 T cells in the blood, many of which share TCR sequences with tumor-infiltrating T cells. The frequency of activated circulating CD8 T cell populations is correlated with the degree of pathologic response within tumors.

Summary

Even a short duration of neoadjuvant immunotherapy can enhance local and systemic tumor-reactive T cell populations. Downstaging induced by neoadjuvant ICB can reduce the extent of surgical resection in this anatomically sensitive location.

Introduction

The activity of tumor-specific T cells is constrained by the PD-1 and CTLA-4 inhibitory receptors that have become major drug targets in oncology. These receptors recruit phosphatases to their cytoplasmic domains which block early steps in T cell activation. Therapeutic antibodies that block the PD-1 or CTLA-4 receptors enhance the anti-tumor function of T cells, including their cytotoxic activity against tumor cells (1,2). These drugs were initially developed in metastatic disease or in an adjuvant setting following surgery (3,4). However, neoadjuvant (pre-surgical) immunotherapy is a rapidly advancing approach that is now supported by a substantial body of data from clinical trials (5–7). An early clinical trial of neoadjuvant ICB with anti-PD-1 monotherapy showed a major pathologic response (≤10% residual viable tumor) in 45% of patients with surgically resectable NSCLC (stage IB-IIIA) (8). Striking clinical activity with neoadjuvant anti-PD-1 has been demonstrated in mismatch repair deficient, locally advanced rectal cancer which demonstrated a clinically complete response in all 12 patients (100%), including no evidence of tumor by MRI, PET imaging or endoscopic evaluation. No progression or recurrence was observed during follow-up (range of 6-25 months), and none of these patients required surgery or chemo-radiotherapy (9).

Two phase 3 clinical trials of neoadjuvant anti-PD-1 and chemotherapy have provided definitive evidence for this therapeutic strategy and FDA approvals. In the CheckMate 816 phase 3 trial, patients with NSCLC were treated for ~12 weeks with anti-PD-1 plus chemotherapy (or chemotherapy alone) prior to surgical resection (10). Importantly, the rate of complete pathologic response (pCR) increased from 2.2% (chemotherapy) to 24% (combination therapy). Also, event free survival increased from 20.8 to 31.6 months for monotherapy versus combination therapy. A phase 3 clinical trial (KEYNOTE 522) in patients with early triple-negative breast cancer (TNCB) also documented greater therapeutic activity for neoadjuvant anti-PD-1 plus chemotherapy compared to chemotherapy alone (11).

An important question is whether neoadjuvant (pre-surgical) or adjuvant (post-surgical) immunotherapy provides better clinical outcomes. This question has been addressed in stage III-IV melanoma (SWOG S1801 trial) in which patients received a PD-1 mAb in either a neoadjuvant (+ adjuvant) or an adjuvant setting. This study documented higher clinical activity for the neoadjuvant approach, with higher 2-year event-free survival (72% versus 49% of patients) for neoadjuvant versus adjuvant ICB (12). This question is being further investigated in a phase 3 clinical trial in metastatic melanoma.

At a conceptual level, neoadjuvant ICB maintains the anatomical connection between the tumor and its draining lymph nodes which is essential for the cancer – immunity cycle in which dendritic cells carry tumor antigens to lymph nodes where they activate T cells (13–15). Also, immune function is not compromised by prior therapies. Mechanistic studies comparing pre-treatment biopsies and surgical specimens have documented striking activation of T cells in a subset of patients, both locally within tumors and in the circulation (16). Such circulating T cells may have been activated within tumor-draining lymph nodes. Enhanced local immunity can account for major or complete pathologic responses, while enhanced systemic immunity may provide protection against distant micro-metastases (5,17) (Figure 1). Clinical to pathologic downstaging can facilitate surgical resection in a subset of patients which is important in anatomically sensitive locations such vital structures of the head and neck area (18). Assessment of pathologic response could also facilitate treatment decisions following surgery, including potential treatment de-escalation or escalation. This review will discuss recent advances in the development of neoadjuvant ICB in head and neck cancer.

Figure 1. Neoadjuvant immune checkpoint blockade enhances local and systemic tumor immunity.

The neoadjuvant approach preserves the connection between the tumor and its draining lymph nodes, allowing migration of DCs from the tumor to its draining lymph nodes where they activate blood T cells. A wave of such activated circulating T cells can be detected at early on-treatment timepoints during neoadjuvant ICB, and a large fraction of such T cells share TCR sequences with tumor-infiltrating T cells. These T cells can migrate to the tumor and micro-metastases. Diagram shows TCR clonotypes (red TCR) shared between tumor-infiltrating (blue) and activated circulating (yellow) T cells.

Evidence for clinical activity of neoadjuvant ICB in HNSCC

Results from a large number of clinical trials have recently been reported in HNSCC which evaluated neoadjvuant ICB (19–21). Even though these studies differed in a number of important details, including the duration of neoadjuvant ICB, they provide substantial preliminary evidence for clinical activity of this treatment strategy. These clinical trials reported no delay in planned surgery, and the median time from neoadjuvant treatment initiation to surgery ranged from 16-46 days across these trials. The rate of clinical to pathologic downstaging was reported for four trials and ranged from 19-100% (18,19,22–24). This aspect is highly relevant for the surgical management of HNSCC because downstaging enables less invasive surgery of these anatomically sensitive structures. Similar to trials in other cancer types, pathologic response was an important readout of therapeutic activity (Table 1). Most studies reported a lack of concordance between radiologic and pathologic tumor response, similar to the experience with neoadjuvant ICB in other cancer types.

Table 1.

Summary of neoadjuvant clinical trials in HNSCC

Clinical trial	Uppaluri et al	Schoenfeld et al	Leidner et al	Hanna et al	Oliveira et al
NCT ID (trial name)	NCT02296684	NCT02919683	NCT03247712	NCT03341936	NCT02296684
Reference	18	14	19	20	24
Cancer type	Stage III/IV HPV-unrelated HNSCC	Oral SCC ≥T2 or LN positive	Stage I-III p16+ and stage III/IVA p16− HNSCC	Recurrent HNSCC	Stage III-IVb HNSCC, HPV unrelated
Patients, n	36	29	21	28	29
Treatment arms	Pembrolizumab, 1 dose 200 mg	Arm 1: Nivolumab 2 doses,	Arms 1, 2 and 4: Nivolumab 3 doses, 240 each	Nivolumab 1 dose, 240 mg	Pembrolizumab, 2 doses 200 mg
		3mg/kg; Arm 2: as Arm 1 +	SBRT: Arm 1 8Gy x 5; Arms 2-4 8Gy x 3	Lirilumab 1 dose, 240 mg
		Ipilimumab 1 dose, 1mg/kg
Time neoadjuvant to surgery	16 (13-22) days	19 (7-21) days	44.6 (40-54) days	13 (6-24) days	40 (33-50) days
Grade 3/4 adverse events	2.80%	Arm 1: 14.2%, Arm 2 33%	1.50%	11%	3.30%
Clinical to pathologic downstaging	19%	Arm 1 69%, Arm 2 53%	90%	54%	27.60%
Pathologic response	44.40%	Arm 1 54%, Arm 2 73%	100%	43%	52%
mPR (% of patients)	5.50%	Arm 1 8%, Arm 2 20%	86%	14%	13.80%
cPR (% of patients)	0%	Arm 1 0%, Arm 2 7%	67%	0%	3.40%
1 year overall survival	High-risk disease 83% (n=18)	Arm 1 92% (n=14)	Not reported	85.7% (n=28)	Not reported
	Intermediate/low risk 100% (n=18)	Arm 2 87% (n=15)
1 year disease free survival	Not reported	Not reported	Not reported	55.2% (n=28)	93%

In a phase 2 clinical trial reported by Uppaluri and colleagues, a single dose of a PD-1 mAb (200 mg) was administered to 36 patients with locally advanced HNSCC, followed two to three weeks later by surgery. Pathologic response was quantified as the proportion of the resection bed with tumor necrosis, keratinous debris and giant cells/histiocytes. A pathologic response of ≥50% was observed in 22% of patients, and a pathologic response of 10-49% in another 22% of patients. Among patients with high-risk pathology, one-year progression free survival was 83.3% (22) (Table 1). At the ASCO 2021 meeting, the same group presented results from a 29-patient phase 2 single arm trial in which HPV-negative patients with HNSCC received two neoadjuvant doses of a PD-1 mAb. A pathologic response of ≥50% was observed in 44% of patients, with a complete pathologic response (pCR) in 4 patients (25). It is possible that the deeper treatment effect in this study was observed due to the longer time window between treatment initiation and surgery.

A randomized phase 2 clinical trial reported by Schoenfeld and colleagues in patients with oral cavity squamous cell carcinoma compared neoadjuvant treatment with PD-1 or PD-1 plus CTLA-4 mAbs (18) (Table 1). Patients received two doses of a PD-1 mAb (3 mg/kg), but only a single low dose of a CTLA-4 mAb (1 mg/kg) to reduce the risk of immune related adverse events. The median time from treatment initiation to surgery was 19 days (range 7-21 days). Several readouts provided evidence for clinical activity, including volumetric response (50% and 53% of patients, mono versus combination therapy) and pathologic downstaging (53 and 69% of patients, mono versus combination therapy). A major/complete pathologic response was observed in 4 patients (1 patient in monotherapy arm, 3 patients in combination therapy arm). One-year progression-free survival rates were 85% and 89% (mono versus combination therapy), even though 62% of patients had stage IVA disease prior to treatment (18). A substantial number of additional clinical trials have been reported with a neoadjuvant approach involving ICB combined with radiation or chemotherapy (19,20).

Importantly, the neoadjuvant treatment paradigm is now being investigated in a phase 3, randomized clinical trial. This trial evaluates neoadjuvant anti-PD-1 in patients with newly diagnosed stage III/IVA resectable, but locoregionally advanced HNSCC. Following surgery, patients receive standard of care radiotherapy, with or without cisplatin. The primary hypothesis is that this regimen improves event-free survival compared to radiotherapy (with or without cisplatin) following surgery (26).

Local immune activation during neoadjuvant ICB in HNSCC

The neoadjuvant approach enables in-depth investigation of dynamic therapy-induced cellular and molecular changes in the tumor microenvironment. In the study by Luoma and colleagues, scRNA-seq was combined with deep T cell receptor (TCR) repertoire sequencing to identify the tumor-infiltrating T cell populations that responded to treatment by clonal expansion (16). The analyzed samples originated from the clinical trial reported by Schoenfeld and colleagues which investigated presurgical PD-1 monotherapy and PD-1 plus CTLA-4 combination therapy (18). The short interval between initiation of neoadjuvant ICB and surgery (median of 19 days) enabled investigation of early changes in the TME.

Use of the TCR as a molecular barcode enabled identification of treatment-expanded T cell clones which were strongly enriched in two clusters of CD8 T cells that expressed markers of tissue-resident memory T cells (Trm): ZNF683 (encoding the HOBIT transcription factor critical for a tissue residency program) and ITGAE (encoding the CD103 surface marker of Trm) (27) (Figure 2). Treatment-expanded T cells also overexpressed cytotoxicity genes (including GZMB and GNLY) and CXCL13 (encoding a chemokine that recruits B cells). One of these two clusters also expressed a cell cycle signature, consistent with significant clonal expansion during treatment. Flow cytometric analysis of pre- and on-treatment samples from a patient with a pathologic response identified a large population of CD8 T cells that co-expressed the CD103 and CD69 markers that are characteristic for Trm. These CD103+ CD69+ CD8 T cells expressed higher levels of granzyme B and PD-1 compared to other CD8 T cells, consistent with the scRNA-seq data. The majority of treatment-expanded TCRs (~60%) were also detected in pre-treatment tumors, but a substantial fraction (~40%) represented ‘emergent’ TCRs detected only in on-treatment but not pre-treatment specimens. T cell clones in pre-treatment tumors that expanded during treatment were almost exclusively localized to the CD8 Trm clusters described above. These data indicate that CD8 T cells with a tissue residency program are poised to respond to neoadjuvant ICB.

Figure 2. T cells with a tissue residency program are early responders to neoadjuvant ICB.

CD8 T cells with a tissue residency program preferentially respond to neoadjuvant immune checkpoint blockade by clonal expansion. Such T cells are characterized by expression of the CD103 integrin receptor (ITGAE gene), the HOBIT transcription factor (ZNF683 gene) and the granzyme B cytotoxicity protein (GZMB gene). The density of such T cells in pre-treatment biopsies is associated with pathologic response to ICB, and CD103+ CD8 T cells preferentially infiltrate tumor areas compared to other T cell populations.

Treatment induced clonal expansion was also observed in CD4 T cell populations, and responding T cells were localized to clusters of CXCL13+ and proliferating T cells. Furthermore, a shared gene program was observed for treatment expanded CD8 and CD4 T cell populations (compared to non-expanded T cells) which included CXCL13, IFNG, LAG3 and GZMA. The target antigens could be identified for seven treatment-expanded TCRs. Two independent TCRs from a patient with a pathologic response recognized the MAGE-A1 tumor antigen, and CD8 T cells expressing these TCRs killed cancer cell lines with endogenous expression of MAGE-A1. Importantly, T cells with defined tumor antigen specificity mapped back to the two T cell clusters representing CD8 T cells with a tissue residency program.

A recent study by Oliveira et al analyzed T cell populations in HNSCC patients treated with neoadjuvant anti-PD-1 (two cycles, surgery after a median of 40 days) (28) (Table 1). Single cell analysis of matched pre- and post-treatment biopsies from six patients showed that responding tumors had clonally expanded CD8 T cells with a tissue resident memory program characterized by ZNF683 (encoding HOBIT) expression and high cytotoxic potential. Pathologic response was associated with a high density of CD103+ PD-1+ CD8+ T cells infiltrating pre-treatment lesions, whereas non-responding tumors demonstrated a relative paucity of such T cells at baseline (Figure 2). This study concluded that PD-1 blockade unleashes the cytotoxic potential of tumor-infiltrating T cells with a Trm program (28).

These two studies highlight the importance of CD8 T cells with a Trm program in pathologic response to neoadjuvant ICB in HNSCC. The first study demonstrates that these T cells represent early responders based on clonal expansion and provides a window into the early changes during neoadjuvant ICB (surgery after a median of 19 days) (16). The second study shows that a higher density of such CD103+ PD-1+ CD8+ T cells at baseline is associated with pathologic response (28). These data thus suggest that tumor-infiltrating CD8 T cells with a Trm and cytotoxicity program are held in check by PD-1 (and other inhibitory receptors), and that their cytotoxic activity can be unleashed by neoadjuvant ICB.

Several recent studies have also identified Trm as a major T cell population implicated in the immune-related adverse events (irAEs) induced by ICB. Single cell analysis of immune cell populations within colitis lesions demonstrated a striking accumulation of CD8 T cells with a highly activated, cytotoxic state. Tracking of TCR clonotypes revealed that the majority of these cytotoxic effector CD8 T cells shared the same TCR sequences with a cluster of tissue-resident memory T cells, thus implicating Trm activation in the biology of irAEs (29). These conclusions were confirmed in two other studies that investigated irAEs in the colon and the skin (30,31).

Systemic immune activation during neoadjuvant ICB

While most studies have investigated local immune activation within tumors, several studies have highlighted the early emergence of activated, proliferating T cells in the systemic circulation (32–34). Enhanced systemic immunity may confer protection against distant micro-metastases because activated T cells upregulate homing molecules that enable entry into peripheral tissues (Figure 1). Luoma et al performed systematic scRNA-seq analysis of PBMCs at three timepoints, including an early 2-week on-treatment timepoint, in 27 patients treated with neoadjuvant PD-1 or PD-1 plus CTLA-4 ICB (16). Among blood CD8 T cells, only a single cluster increased in frequency at the 2-week on-treatment timepoint, and this cluster represented activated (HLA-DR, CD38), proliferating (MKI67) T cells. Importantly, blood and tumor TCRs that clonally expanded during treatment mapped to this CD8 T cell cluster. The connection between activated circulating CD8 T cells and tumor-infiltrating T cells was further substantiated by sorting of activated CD8 T cells (CD38+, HLA-DR+) from blood samples of four patients for scRNA-seq analysis. The majority of TCR sequences from these activated circulating T cells were also identified among tumor-infiltrating T cells. Treatment-responsive TCR clonotypes peaked in the blood at the 2-week on-treatment timepoint but contracted to low/undetectable levels at the 10 to 12-week timepoint, potentially due to homing of these T cells to tissues (16). It is likely that these T cells were activated in tumor-draining lymph nodes during neoadjuvant ICB and entered the blood on their route to the tumor and other tissues. The anatomical integrity of the tumor - lymph node connection may be an important feature of neoadjuvant ICB because it allows dendritic cells from the tumor to activate T cells in tumor-draining lymph nodes (Figure 1).

Blood-based biomarkers of pathologic response within HNSCC

The identification of T cell clones that expand in both tumor and blood during neoadjuvant ICB also suggests that activated blood T cells could serve as a biomarker of pathologic response. This question was addressed by flow cytometric analysis of blood T cells from 28 patients across pre- and on-treatment timepoints. Activated blood CD4 and CD8 T cells (HLA-DR+ CD38+) expanded substantially at the earliest, two-week on-treatment timepoint, and a large fraction of these activated CD8 T cells was also Ki67+, indicative of a proliferative state (16).

An important question is whether the activation state of peripheral blood T cells could also serve as a pre-treatment biomarker of later pathologic response. Indeed, the frequency of activated CD8 T cells (HLA-DR+ CD38+) was substantially higher at the pre-treatment timepoint in patients with a pathologic response of >50%. PD-1 expression was also upregulated by activated T cells, but the frequency of PD-1+ T cells was not higher in patients with a tumor pathologic response. However, PD-1+ circulating T cells could be subdivided based on expression of KLRG1, a marker of terminal T cell differentiation (35). Indeed, the frequency of PD-1+ yet KLRG1- CD8 T cells at the pre-treatment timepoint was associated with the depth of pathologic response, potentially because KLRG1 negative T cells have greater proliferative potential (16). These blood biomarkers could therefore be used to identify patients who are more likely to respond to neoadjuvant ICB, but this concept needs to be validated in prospective studies.

Conclusions

Neoadjuvant ICB is rapidly advancing as an important treatment paradigm in many different cancers. The available clinical trial data provide substantial evidence for clinical activity in HNSCC but need to be validated in randomized phase 3 clinical trials. Many clinical trials are also ongoing to develop combinatorial strategies based on these promising data. Mechanistic studies have highlighted the importance of CD8 T cells with tissue residency and cytotoxicity programs. CD8 T cells with such molecular programs rapidly expand during neoadjuvant ICB, and the density of such T cells in pre-treatment biopsies correlates with pathologic response. Neoadjuvant ICB maintains the integrity of the functionally important connection between the tumor and its draining lymph nodes which likely accounts for the wave of activated T cells detected in the circulation early during treatment. Thus, neoadjuvant ICB not only enhances local but also systemic tumor immunity.

Key points.

• A series of clinical trials have provided substantial evidence for clinical activity of neoadjuvant ICB in HNSCC, and a randomized phase 3 clinical trial is ongoing to validate this conclusion.

• The extent of pathological response is emerging as a major indicator of treatment efficacy in HNSCC and other cancer types.

• Tumor-infiltrating CD8 T cells that respond early during treatment are characterized by tissue residency (Trm) and cytotoxicity programs, and the density of such T cells in pre-treatment biopsies correlates with the degree of pathologic response.

• Activated proliferating T cells are detected in the blood at an early 2-week on-treatment timepoint; a large fraction of these activated cells share T cell receptor sequences with tumor-infiltrating T cells.

• The presence of activated circulating T cells correlates with the degree of pathologic response in the tumor.