Head & Neck Cancer (Part 2): Treatment, NPC, CUP and Recurrent Disease

Continued from: Head & Neck Cancer (Part 1): Biology, Workup and Staging

Treatment: general principles by stage and HPV status

• HPV− stage I, II, low-bulk III (T1/T2, N0 or N1) or HPV+ stage I (nonbulky): ~20 to 30% of HNSCC; single-modality surgery OR RT is curative (cure rate 50 to 100%). Base postoperative treatment on pathology: positive margins warrant re-resection or chemoRT, ENE warrants chemoRT; other adverse features (e.g., pT3-4, multiple nodes, LVI/PNI) warrant postoperative RT without chemo. RT gives excellent control and good voice outcome; conventionally fractionated definitive RT runs about 7 weeks (70 Gy/35 fx) with late fibrosis, but early glottic cancers use shorter hypofractionated courses.
• Glottic T1: transoral laser microsurgery (TLM) or RT (similar OS); laryngeal-preservation focus.
• HPV− stage III and IVa/IVb, or HPV+ stage I (bulky), II, III: ~60% of HNSCC; cure depends on resectability. If resectable → surgery then adjuvant RT ± chemo by pathology (chemoRT ↑ organ preservation). If unresectable → definitive chemoRT (cure 10 to 65%). When primary chemoRT is used, surgery is reserved for persistent or resectable recurrent disease. For resectable oral cavity, primary surgery with reconstruction plus risk-stratified postoperative RT ± chemo is the mainstay.
• Neoadjuvant plus adjuvant pembrolizumab is FDA approved for resectable stage III to IVA HNSCC with CPS ≥1 (KEYNOTE-689, Jun 2025); other neoadjuvant PD-1 strategies remain investigational.

Surgery

• Transoral robotic surgery (TORS): ↑ visualization, ↓ morbidity. Advanced p16+ oropharynx has similar OS with definitive surgery or definitive chemoRT.
• Neck dissection types: radical/comprehensive (en bloc removal of all 5 LN levels plus SCM, internal jugular vein, spinal accessory nerve); modified radical (any/all of those structures preserved); selective (specific levels only, e.g., larynx → bilateral levels II to IV).
• Elective neck dissection in early oral cavity (cT1 to T2, N0) improved 3-yr OS (80% vs 67.5%) and 3-yr DFS (69.5% vs 45.9%) vs therapeutic dissection at relapse (PMID 26027881). SLN biopsy is an option in early clinically N0 oral cavity HNSCC.
• Reconstruction: free flaps (skin, muscle, bone), obturators/prosthetics enable large resections with good function.

Radiation therapy

• HPV+ → ↑↑ radiosensitivity. RT alone can treat early-stage disease.
• Adjuvant RT: start no later than 6 weeks post-op; usual dose ~60 Gy; boost positive margin or nodal ENE to 66 Gy; uninvolved/nondissected nodes 50 to 56 Gy. ECOG-ACRIN 3311 reduced dose from 60 to 50 Gy for ENE <1 mm, clear/close margins, 2 to 4 LN+. Dental clearance before RT.
• Acute toxicity: dry mouth ~50%, mucositis 20 to 40%, desquamation, pain ~15%, odynophagia, dysgeusia, thick mucus. Grade ≥3 acute ~70%, chronic ~40%. 1-yr feeding tube: 20% HPV− vs 5% HPV+. Xerostomia is a late effect.
• IMRT: more dose to tumor, less to normal tissue; spares constrictors and salivary glands (↓ xerostomia) but ↑ fatigue vs conventional RT. Proton beam: sharp dose falloff beyond tumor; phase 2 data in base-of-skull, periorbital, nasopharynx, and re-irradiation.
• Fractionation: standard = once-daily 2 Gy/d to ≥70 Gy (primary/gross nodes), 50 to 60 Gy to at-risk nodes. Altered fractionation (umbrella term: hyperfractionation and/or acceleration) improves locoregional control; hyperfractionation = twice daily; accelerated = 6 fractions/week >6 h apart. RTOG 0129: cis ×2 with accelerated-boost RT vs cis ×3 with standard RT gave no OS/PFS difference.

Chemotherapy

• Cisplatin is the best single agent; others: 5-FU, taxanes (paclitaxel, docetaxel), methotrexate, cetuximab. Induction ORR 60 to 90% (clinical CR 20 to 50%); recurrent-disease ORR 30 to 40% (CR rare).
• Four settings: induction/neoadjuvant, definitive chemoRT (unresectable), adjuvant chemoRT (resectable, after surgery), and organ preservation.

• TPF (taxane + cisplatin 100 mg/m² + 5-FU 1,000 mg/m²/day ×4 d continuous infusion, TAX 324 dosing) is SOC induction, superior to PF (cisplatin + 5-FU) across GORTEC 2000-01, TAX 323, TAX 324 and a 5-trial meta-analysis (1,772 pts; PMIDs 21233014, 17960012, 26681800). GORTEC 2000-01 (larynx/hypopharynx): 3-yr laryngeal preservation 70.3% vs 57.5%, ORR 80% vs 59.2%. TAX 323 (unresectable): mPFS 11 vs 8.2 mo, CR 33% vs 19.9%. TAX 324: 5-yr OS 52% vs 42%, mPFS 38.1 vs 13.2 mo.
• Induction TPF before chemoRT shows no OS benefit; NPC is the one setting with randomized support for induction before chemoRT. Induction may be used in selected larynx-preservation strategies but without a proven OS advantage. Consider induction to rapidly shrink bulky neck disease, not to improve OS.

• SOC for locally advanced unresectable HNSCC and for organ preservation of larynx/oropharynx and locoregionally advanced NPC. ChemoRT gives ~8% 5-yr OS benefit vs RT alone (HR 0.81; MACH-NC meta-analysis, Pignon, PMID 10768432).
• Cisplatin = gold standard: 100 mg/m² q3wk ×3 (days 1, 22, 43) or 40 mg/m² weekly; need ≥200 mg/m² cumulative for similar efficacy. Cumulative cisplatin ≥200 mg/m² drives efficacy; high-dose q3wk (100 mg/m²) has the strongest definitive evidence, weekly 40 mg/m² is an accepted alternative. Laryngeal cancers were included in major chemoRT trials (e.g., RTOG 91-11); some organ-preservation trials excluded high-volume T4. Multiagent chemoRT (e.g., cisplatin + 5-FU) offers no advantage over single-agent cisplatin (87-trial, >16,000-pt analysis, PMID 19446902). Toxicity: hearing loss, N/V, renal dysfunction, peripheral neuropathy.
• Cetuximab + RT (Bonner trial): better than RT alone (mOS 49.0 vs 29.3 mo, HR 0.73; ↑ OS with grade 2 acneiform rash), but inferior to cisplatin + RT (RTOG 1016, TROG 12.01, both in HPV+/p16+ oropharyngeal cancer). Only indication with RT: cisplatin contraindication (severe hearing loss, CrCl <50 mL/min, severe neuropathy) or patient preference. Cetuximab has no effect on distant metastases. Do NOT add cetuximab to cis-RT (RTOG 0522: no PFS/OS benefit).

• RTOG 9501 (high-dose cis + RT vs RT alone): high-risk = R1, ENE, or ≥2 LN+; PFS benefit and trend to OS in the R1/ENE+ subgroup. EORTC 22931: high-risk = R1, ENE, PNI, LVI (and, for oral/oropharynx, level IV or V LN+); 5-yr PFS 47% vs 36%, 5-yr OS 53% vs 40% favoring chemoRT.
• Pooled analysis: microscopically positive margin (R1) or ENE is an absolute indication for adjuvant chemoRT.
• Kiyota (Japanese trial, R1/ENE+): weekly cisplatin 40 mg/m² + RT vs cisplatin 100 mg/m² q3wk + RT gave 3-yr OS 71.6% vs 59.1% (HR 0.69) with less grade ≥3 neutropenia and hearing impairment, supporting weekly cisplatin as a noninferior adjuvant option.

HPV+ oropharyngeal cancer: deintensification

• Concept: HPV+ disease is more curable; trials explore reduced-dose RT, chemo omission, or TORS.
• Cetuximab is INFERIOR to cisplatin (RTOG 1016); cisplatin + RT remains standard. Deintensification trials: ECOG E3311 (phase 2, 2022): 50 Gy after TORS in intermediate-risk pts gave 2-yr PFS ~95% (vs ~96% with 60 Gy), promising but not SOC; NRG-HN005 (reported): reduced-dose arms failed noninferiority; 2-yr PFS 98.1% (70 Gy + cisplatin) vs 88.6% (60 Gy + cisplatin) vs 90.3% (60 Gy + nivolumab). Standard-dose cisplatin + RT remains SOC. Off trial → standard cisplatin + RT.

• Indication: resectable locally advanced (stage III to IVA) HNSCC with PD-L1 CPS ≥1.
• Regimen: neoadjuvant pembro ×2 → surgery → adjuvant pembro + RT ± cisplatin → adjuvant pembro alone.
• Efficacy: mEFS 59.7 vs 29.6 mo (HR 0.70) in CPS ≥1.
• FDA approval: Jun 12, 2025. First perioperative IO approval in HNSCC.

Organ preservation (larynx, hypopharynx, oropharynx)

• For locally advanced larynx/hypopharynx/oropharynx, chemoRT > RT alone (↑ local control, ↑ AEs). Monitor mucositis, weight loss, cytopenias, dysphagia, feeding tube need. Salvage surgery when needed. Post-treatment PET at ~12 weeks in LN+ disease: At ~12 weeks, intense residual nodal FDG uptake warrants neck dissection; mild/equivocal uptake or enlarged metabolically negative nodes can be surveilled with repeat imaging. Size alone does not mandate dissection after a complete metabolic response.
• Larynx/hypopharynx: cisplatin 100 mg/m² days 1, 22, 43 + RT is SOC for larynx preservation; surgery reserved for persistent/recurrent disease. Large T4a tumors with poor pretreatment laryngeal function → total laryngectomy (no randomized support for organ preservation there).
• VALCSG: induction cis + 5-FU then RT vs laryngectomy (resectable T2 to T4 larynx) had same OS; laryngectomy avoided in ~2/3. T4 and N2 predicted failure (56% of T4 eventually needed laryngectomy). EORTC (hypopharynx T2 to T4): same OS, 5-yr successful larynx preservation ~35%.
• RTOG 91-11 (547 pts, T2 to low-volume T4 larynx): chemoRT vs induction-chemo then RT vs RT alone. Larynx preservation 88% vs 75% vs 70% (10-yr LP 81% vs 68% vs 64%); locoregional control 78% vs 61% vs 56%; OS not significantly different (10-yr 27.5% vs 38.8% vs 31.5%, a nonsignificant trend favoring induction). Conclusion: chemoRT is SOC for organ preservation; advanced T4 or nonfunctional larynx → total laryngectomy.
• Oropharynx: cisplatin 100 mg/m² + RT 70 Gy is SOC; concurrent weekly cisplatin + RT if not a high-dose candidate. Carboplatin (70 mg/m²/d ×4) + 5-FU + RT is an alternative to high-dose cisplatin (↑ toxicity, esp. cervical fibrosis). Early-stage low-volume (HPV− T1-T2 N0-N1, or HPV+ T1-T2 N0-N1 with N1 a single node <3 cm): RT alone or surgery, avoid radical glossectomy/chemoRT.

Nasopharyngeal carcinoma (NPC): distinct disease

• Histology: WHO type I (keratinizing, more common in nonendemic/Western, smoking-associated, least EBV-linked but EBV can be present; HPV+ is a minority subset) vs types II/III (nonkeratinizing, >90% EBV, endemic in Asia/Africa, higher distant-met propensity, respond better to chemoRT). EBV-negative NPC is often HPV+.
• EBV DNA (plasma): the most significant prognostic biomarker; used for screening (endemic areas), staging, and surveillance. Adding adjuvant chemo did not improve relapse-free survival in those with detectable post-RT EBV DNA.

• Stage II: high-risk (cervical LN >3 cm, level IV/VB nodes, ENE+, pretreatment plasma EBV DNA >4,000 copies/mL) → concurrent chemoRT (5-yr OS 95% vs 86% vs RT alone, benefit mostly in T2N1); low-risk → RT alone. T2N0 → definitive RT.
• Stage III to IVa: induction chemo then concurrent chemoRT is preferred. Induction gemcitabine 1 g/m² days 1 and 8 + cisplatin 80 mg/m² day 1, q21d ×3 for T3 to T4, N1 to N3 or any T, N2 to N3 or oligometastatic disease (MAC-NPC meta-analysis, 28 RCTs/8,214 pts: induction + chemoRT improved OS, HR 0.75 with taxane, 0.81 without). For concurrent chemoRT, both every-three-week cisplatin 100 mg/m² (most evidence-based) and weekly 40 mg/m² are accepted; choose by patient factors, with carboplatin/oxaliplatin if cisplatin-ineligible.
• T3N0: RT alone can be considered without high-risk features (Tang et al: 3-yr FFS 90.5% RT vs 91.9% chemoRT, noninferior, fewer AEs); strongly consider chemoRT if high-risk. Induction/adjuvant chemo individualized.
• Sequential induction then chemoRT: Sun et al (TPF + chemoRT vs chemoRT) 3-yr FFS 80% vs 72%, 3-yr OS 92% vs 86%; Zhang et al (gem/cis induction then chemoRT) 3-yr RFS 85.3% vs 76.5%, 3-yr OS 94.6% vs 90.3% (more grade 3 to 4 AEs).
• Adjuvant capecitabine (Chen et al, Lancet 2021): metronomic capecitabine ×1 yr improved 3-yr FFS (85.3% vs 75.7%, HR 0.50) in high-risk NPC after concurrent chemoRT. NRG-HN001: EBV-DNA-guided adjuvant strategy (pending).
• Recurrent/metastatic: gem + cis is SOC (category 1; PFS 7.0 vs 5.6 mo vs 5-FU/cisplatin). Add PD-1 blockade: toripalimab (JUPITER-02, PFS 11.7 vs 8.0 mo; FDA approved Oct 2023), camrelizumab (CAPTAIN-1st, PFS 9.7 vs 6.9 mo, HR 0.54), tislelizumab (RATIONALE-309). Penpulimab-kcqx + cisplatin or carboplatin/gemcitabine is FDA approved (Apr 23, 2025) for 1L R/M nonkeratinizing NPC. US-available PD-1 agents (pembrolizumab ORR ~28%, nivolumab ORR ~20%) can be added to gem/cis per NCCN.

Carcinoma of unknown primary (CUP) in the H&N

• FDG-PET identifies the primary in ~30% of SCC-of-unknown-primary cases. FNA of the node first; if negative, repeat FNA ± core biopsy.
• Viral testing of the nodal specimen (HPV and EBV): if positive, tumor is T0 and treated per oropharyngeal (HPV+) or nasopharyngeal (EBV+) protocol. If HPV-negative and EBV-negative and no primary is found after workup, stage as T0 (cervical nodes/unknown primary) with the appropriate N and M, not Tx.
• If SCC with a likely primary above the clavicle → examination under anesthesia with directed biopsies of hypopharynx, base of tongue, nasopharynx, plus bilateral tonsillectomies (rising HPV+ tonsillar cancers). Skin, upper esophagus, and lung are other SCC sources; thyroid/salivary or subclavicular sites suggest adenocarcinoma; consider lymphoma (core biopsy).
• Treatment: surgery or RT for low-bulk single node without ECE; surgery + RT for advanced or ECE+. Mainstay is neck dissection then RT to likely primary sites, with concurrent cisplatin if ECE+. As most are p16+ oropharyngeal, definitive chemoRT is another standard option. Avoid inappropriate open biopsy (seeding).

Incurable recurrent / metastatic disease

• KEYNOTE-048 (882 pts) vs EXTREME (cetuximab + platinum/5-FU): pembro improved OS in CPS ≥20 (14.9 vs 10.7 mo, HR 0.61) and CPS ≥1 (12.3 vs 10.3 mo, HR 0.78); pembro + chemo improved OS in CPS ≥20 (14.7 vs 11.0 mo), CPS ≥1 (13.6 vs 10.4 mo), and overall (HR 0.77). CPS = PD-L1+ cells (tumor, lymphocytes, macrophages) / total tumor cells ×100, by 22C3 pharmDx.
• Choice: CPS ≥20: pembro monotherapy and pembro + platinum/5-FU are both standard; favor monotherapy for indolent, low-burden disease and combination when a rapid or higher-probability response is needed; CPS ≥1 → pembro mono or pembro + platinum/5-FU; CPS <1 (rare) → pembro + platinum/5-FU or EXTREME. FDA approved Jun 2019; 5-yr follow-up confirms durable benefit. KEYNOTE-B10: 1L carbo/paclitaxel + pembro, ORR 49% (BICR), mOS 13.1 mo.
• EXTREME regimen (cetuximab + cisplatin/carboplatin + 5-FU): mOS 10.1 vs 7.4 mo, mPFS 5.6 vs 3.3 mo, ORR 35% vs 20% vs platinum/5-FU; now reserved for IO-ineligible patients.
• Taxane doublets (cisplatin + paclitaxel): similar efficacy to cis/5-FU with less toxicity (ECOG: OS 9 vs 8 mo; more grade 3 to 5 AEs, mucositis, hand-foot, and cardiac ischemia in the 5-FU arm, worse with DPD deficiency).

• Nivolumab (CheckMate-141): mOS 7.5 vs 5.1 mo (HR 0.70), ORR 13.3%, 1-yr OS 36.0% vs 16.6%; PD-L1 not required; HPV+ respond better. FDA approved 2016.
• Pembrolizumab (KEYNOTE-040): mOS 8.4 vs 6.9 mo (HR 0.80), ORR 14.6% vs 10.1%.
• Ipi/nivo (CheckMate-651) vs EXTREME was a negative trial.
• Cetuximab: monotherapy for R/M HNSCC progressing after platinum; platinum/5-FU combination (EXTREME) is for 1L R/M disease. With cisplatin: ORR 26% vs 10%, similar PFS. Monotherapy 2L: ORR ~13%, disease control ~46%; acneiform rash/dry skin >50%, hypomagnesemia, risk of anaphylaxis. After IO progression: cetuximab, methotrexate, taxane, gemcitabine.

Survivorship & supportive care

• Pre-RT dental clearance (extract non-restorable teeth).
• Xerostomia: pilocarpine, amifostine (less used); IMRT spares parotids. Mucositis: magic mouthwash, pain management, nutrition (PEG).
• Hypothyroidism: monitor TSH q6 to 12mo after neck RT, continue long-term (lifelong risk). Carotid stenosis: duplex surveillance after neck RT.
• Smoking/alcohol cessation improves outcomes and reduces second primaries.
• Sinonasal undifferentiated carcinoma (SNUC): aggressive; multimodality; test PD-L1, NUT, IDH.

High-yield H&N pearls

• HPV+ oropharyngeal: better prognosis, separate staging; do NOT deintensify off-trial. p16 IHC positive = ≥70% moderate-to-strong nuclear and cytoplasmic staining (HPV surrogate); used in oropharyngeal SCC and in cervical nodal SCC of unknown primary, with HPV-specific testing per context.
• Cisplatin + RT is the standard concurrent regimen; cetuximab is INFERIOR (RTOG 1016) and not added to cis-RT (RTOG 0522).
• R1 margin or ENE → adjuvant cis + RT (RTOG 9501 / EORTC 22931); weekly cisplatin may be noninferior (Kiyota).
• TPF is SOC induction; induction then chemoRT gives no OS benefit except NPC (larynx preservation: no proven OS gain).
• Larynx preservation (RTOG 91-11): concurrent cis + RT gives best laryngeal preservation; total laryngectomy for extensive T4a (invasion through the outer cortex of thyroid cartilage or beyond the larynx; inner-cortex invasion alone is T3) or a nonfunctional larynx.
• KEYNOTE-048: 1L pembro (± chemo) for CPS ≥1 R/M HNSCC. KEYNOTE-689 (Jun 2025): perioperative pembro for resectable CPS ≥1 stage III to IVA.
• NPC: EBV DNA biomarker; gem/cis ± toripalimab 1L metastatic; separate staging.
• CUP: PET-CT, EUA with directed biopsies plus bilateral tonsillectomy; HPV/EBV testing directs treatment (T0 if viral+).