48.1 Introduction

Although the incidence of oral cavity cancer (OCC) has been decreasing in North America, it remains one of the most commonly diagnosed cancers worldwide. ¹ It is important to understand the pathophysiology of oral cavity carcinoma which predisposes to locoregional recurrence (LRR) in order to optimize the success of primary curative therapy, as well as to undertake successful treatment of LRR. LRR refers to evidence of cancer at the site of primary disease (local) and/or the local draining lymph node basin (regional) after completion of curative treatment. Timely diagnosis of recurrent cancer improves the likelihood of cure; therefore, a high level of clinical suspicion and close follow-up during the high-risk period of follow-up are necessary. This chapter will focus on discussing the epidemiology of recurrent oral cavity cancer (rOCC), risk factors for recurrence, presentation and diagnosis of rOCC, as well as management strategies in this complex disease process.

48.2 Epidemiology: Incidence, Causes, and Clinical Characteristics

The oral cavity is divided into subsites, which include the oral tongue, floor of mouth, hard palate, upper and lower alveolus and gingiva, and buccal mucosa, which is composed of the retromolar trigone, mucosa of the upper and lower lips, cheek mucosa, and the upper and lower buccoalveolar sulci. ^{2 , 3} The current standard of care treatment for OCC is comprised of surgical excision with or without adjuvant radiation or chemoradiation. ^{4 , 5} Indications for adjuvant radiotherapy include close or positive resection margins, extracapsular nodal extension (ENE), perineural invasion (PNI), N2 or N3 nodal disease, and vascular invasion, ^{6 , 7} whereas ENE and positive margins are indications for the addition of chemotherapy to adjuvant treatment. ^{8 , 9} Over 80% of patients who present with early stage (T1-T2, N0-N1) OCCs will experience cure with standard therapy. However, nearly half of all patients with OCCs present with locoregionally advanced disease (T3-T4, N2-N3), ^{10 , 11} which carries an increased likelihood of developing recurrent disease. In this population, recurrence rates have been reported between 20 and 50%, ^{12 – 14} with local recurrences occurring more frequently than regional recurrences (65 vs 25%, respectively). ¹⁴

Although presence of nodal metastases at the time of diagnosis is the most significant prognostic factor for regional recurrence, ¹⁵ other factors affecting the propensity for oral cancer to recur include tumor thickness, margin status, histological grade, and extracapsular spread of metastatic nodal disease. ^{15 – 18} OCCs have a propensity for bilateral nodal disease, which occurs due to certain OCC subsites having crossover lymphatic drainage, as well as absence of barriers to spread, predisposing to direct invasion to contralateral tissue. ^{19 , 20} Therefore, adequate primary treatment of the draining lymph node basin needs to be undertaken at the time of primary treatment to prevent local recurrences. Preservation of the submandibular gland can be considered to improve patients’ quality of life by maintaining saliva production which is at risk of decreasing with postoperative radiation treatments. However, careful patient selection is important to avoid an increased risk of LRR. A study by Lanzer et al ²¹ found that patients with oral cancers originating in the tongue or floor of mouth had higher rates of LRR when the ipsilateral submandibular gland was not resected as part of the neck dissection. Lymph node recurrences were observed in 28.5% of the patients whose submandibular gland was preserved compared with 2.2% where the gland was removed as part of the neck dissection (p < 0.001). Elective neck dissections play an important role in the management of OCC with N0 disease, since occult nodal metastases occur in over 20 to 30% of OCC patients. ^{22 , 23} The incidence of contralateral occult metastasis has been reported to be around 10% and is statistically significantly predicted by presence of T3 or T4 primary disease, extension of the primary cancer across the midline, presence of positive ipsilateral lymph nodes, poor differentiation on histologic grading, and positive surgical margins. ^{24 , 25} A study by Capote-Moreno et al ²⁵ showed that contralateral nodal disease is associated with poor locoregional control and survival outcomes, with 5-year, cause-specific survival rate of 41.2% for those with contralateral nodal metastases compared with 70% in those without contralateral lymph nodes (p < 0.01). ²⁵ The study also found that contralateral lymphadenopathy did not occur in primaries originating in the retromolar area and the hard palate, were uncommon in buccal mucosa and gum cancers (3.2 and 5.1%, respectively), and occurred most commonly in cancers of the floor of mouth, followed by tongue cancers (11 and 7.2%, respectively). ²⁵

There has been a long controversy regarding the pathological features of oral malignancies which should trigger neck dissection in order to prevent regional recurrences, especially in the context of small tumors (T1 or T2). Depth of invasion of the OCC primary has been known to be associated with the development of nodal disease for nearly four decades, with landmark studies published by Spiro et al from Memorial Sloan Kettering as early as 1986. ^{26 , 27} These studies identified that tumors with a depth greater than 2 mm were associated with a higher rate of treatment failure, and the depth of invasion cutoffs better predicted overall survival than conventional T stage based on tumor size. These studies also emphasized the importance of performing elective neck dissections while nodal metastases remained clinically occult to improve patient outcomes. Subsequent studies continued to show an association between tumor thickness and LRR, but cutoffs for tumor thickness as an indication for elective neck dissections have varied between 2 and 10 mm, ^{17 , 26 , 28 – 31} and little consensus exists to guide clinical care. This may be due in part to variability in the patterns of tumor growth, as tumor thickness has not reliably reflected the differences between endophytic and exophytic invasion patterns ¹⁷ (see ▶ Fig. 47.5a-c). More recently, tumor depth as determined in relation to the basement membrane has been shown to closely and more reliably correlate with disease-free survival. In fact, a recent study by Shim et al ¹⁸ found that 5 years disease-free survival in patients with T1 and T2 and N0 or N1 disease was 66% for tumors with 5 mm or greater depth of invasion, compared with 92% for tumors with less than 5 mm depth of invasion (p = 0.013). As a result of these findings, the 8th edition of the American Joint Committee on Cancer (AJCC) staging, which has been adopted since January 1, 2018, has now included depth of invasion as a criterion for T staging in order to better reflect prognostic features of cancers into the current staging system. ³²

Tumor resection margin status is also particularly relevant in predicting the likelihood of recurrences. In fact, Loree and Strong ³³ first demonstrated this in 1990, showing that a positive margin at the time of surgical excision doubled the rate of recurrence. In this study, patients with negative margins were found to have a recurrence rate of 18% compared with 36% for those with positive margins. Of those with positive margins, patients with in-situ carcinoma at the margins had the highest recurrence rate (44%), followed by microinvasion (36%), margins within 0.5 cm or close margins (35%), and premalignant changes at the margins (33%). They also found that positive margins were more likely to be found in primaries of the buccal mucosa (58%), followed by upper and lower gingiva (42 and 39%, respectively), whereas primaries of the retromolar trigone, tongue, and floor of mouth had the lowest rates of positive margins (24, 27, and 31%, respectively). Moreover, increasing T stage was positively correlated with positive margins and likelihood of recurrence. ³³ Since then, studies have confirmed similar findings for reduced overall survival and increased recurrence rate for patients with pathologically positive surgical margins. ³⁴

Another feature of primary OCC that predicts poor outcomes following primary treatment and higher incidence of recurrence is the subsite of the primary cancer. Although relatively common in Southeast Asia due to use of betel quid, ³ OCCs originating in the buccal mucosa comprises roughly 10% of oral cancers in the Western World; yet they are associated with significantly decreased survival compared to other oral cavity subsites. ³⁵ In fact, the 2- and 5-year disease-specific survival (DSS) were 68.0 and 57.3% compared with 73.4 and 63.8% for other oral cavity subsites, and the overall survival (OS) was 60.49 and 41% for buccal cancers compared with 66.8 and 51.1% for other OCC subsites. ³⁵ Compared to the other oral cancer subsite counterparts, buccal cancers are more likely to present with nodal metastases, and at more advanced local T stage, and affect older patients more than other OCCs. ^{35 – 37} In fact, occult metastases occur in 32% of buccal cancers, and 43% of buccal cancers develop nodal metastases. ³⁸ Moreover, nodal metastases occur in over 40% of T1 and T2 disease. ³⁸ It is unclear why buccal cancers present at more advanced stages than OCCs of other subsites, but it is believed that identification of premalignant or early disease may be more challenging than OCCs of other subsites. ³⁵

Advances in genetic sequencing have allowed us to identify genetic factors that predict aggressive clinicopathologic behavior of oral cavity tumors. For instance, the number of copy number alterations (or CNAs) such as losses or gains of certain chromosomal regions can predict clinical outcomes better, where smaller numbers of CNAs correlate with a more favorable prognosis. ³⁹ This genetic subset of oral cavity malignancies arises from genetic mutations and exhibits a three-gene pattern of an activating mutation of HRAS, an inactivating mutation of CASP8, and a wild-type TP53. These changes are thought to result from chromosome 7 amplifications which include an EGFR locus. Conversely, inactivation or loss of the TGFBR2 gene has also been identified in certain OCC, and is associated with poor prognostic outcomes such as increased risk of metastasis. ⁴⁰

48.3 Clinical Presentation

Over 80% of recurrences occur within the first 2 years of treatment. ^{41 – 44} After completion of curative treatment of primary oral cavity malignancies, ongoing follow-up is maintained and early identification of recurrences is important to ensure timely salvage treatment if feasible. A thorough history and physical examination are critical for identifying recurrent disease. Signs and symptoms of recurrent locoregional disease can be similar to those of the original presentation of malignancy and can also be confused for side effects of surgical or adjuvant treatment; ⁴⁵ therefore, maintaining high clinical suspicion for recurrence is essential for identification of recurrent disease. Most patients present with local rather than regional recurrences, ¹⁶ with signs and symptoms including new or worsening pain, nonhealing oral ulcer, bleeding, dysphagia, odynophagia, trismus, decreased tongue mobility, tooth loss, and numbness of the tongue, oral cavity, lips, and/or face. ^{44 , 46} An example of this diagnostic challenge is the difficult to treat sequela of osteoradionecrosis caused by radiotherapy to the mandible, which causes a range of symptoms such as pain, halitosis, pathological fracture of the mandible, and ensuing weight loss. A retrospective case series found that 3% of resected osteoradionecrosis specimens contained unexpected evidence of malignancy. ⁴⁷

48.4 Diagnosis and Evaluation

According to the 8th edition of AJCC, recurrent cancer is defined by the presence of cancer after a disease-free period ^{2 , 48} following treatment with curative intent, and is not to be mistaken with residual or persistent disease. Although the AJCC does not suggest an explicit cutoff for the duration of a “disease-free period,” to prevent confusion between persistent and recurrent disease, experts have suggested 3 to 6 months as the lag time between completion of treatment and designation of cancer as recurrent. ^{13 , 14} According to the 2018 National Comprehensive Cancer Network (NCCN) guidelines for treatment of head and neck cancers, follow-up for oral cancers include a complete head and neck examination with fiberoptic or mirror laryngoscopy every 1-3 months for the first year, every 2-6 months for the second year, and then every 4-8 months until 5 years posttreatment. ⁴⁹ After 5 years, screening is recommended on a yearly basis. Repeat imaging is recommended within 6 months of completion of therapy and subsequently for worrisome signs or symptoms or to evaluate areas that are difficult to examine. ⁵⁰ The type of imaging used for follow-up assessment should remain consistent with preoperative imaging for ease of comparison, and typically consists of computed tomography (CT) and/or magnetic resonance imaging (MRI), ideally contrast-enhanced. ⁵¹ Ultrasonography can also play a role in identifying cervical lymphadenopathy, but contrast-enhanced CT (CECT) imaging remains superior due to its increased diagnostic yield when compared to ultrasound and/or MRI. ^{52 , 53} The local effects of surgery and adjuvant radiotherapy and chemotherapy lead to many changes in tissue properties, making it difficult to differentiate the nature of imaging findings, which can be caused by reactive inflammatory processes, scarring, and edema or may represent recurrent disease. ^{54 , 55} An imaging modality that can help differentiate posttreatment changes is positron emission tomography (PET)-CT, which has a high sensitivity and specificity for recurrence. ⁵⁶ Nonetheless, risk stratification for likelihood of LRR based on such imaging has long been challenging and imprecise. In an effort to overcome this challenge, a standardized reporting template ⁵⁷ was created, similar to those already existing for malignancies of the breast and prostate. This reporting template for CECT and PET-CT imaging named the Neck Imaging Reporting and Data System (NI-RADS) stratifies patients into four risk categories and proposes management strategies based on NI-RADS Category (see ▶ Table 48.1). The classification system was accurate for identifying primary recurrences, nodal recurrences, and combined sites (area under the curve or AUC, respectively, 0.787, 0.712, and 0.756; p < 0.001), and was slightly more accurate for CECT than for PET-CT (AUC 0.779 vs. 0.709).

Ultimately, assessment of rOCC should aim at differentiating and stratifying patients into those amenable to salvage curative therapy and those who are better geared toward palliative therapy. Ability to resect disease is an important factor in making this distinction and is usually defined by encasement of the carotid artery, invasion of prevertebral fascia, and invasion of intracranial structures or mediastinal vessels. Other relevant factors that can affect the decision of pursuing further treatment include the patient’s willingness or ability to complete further treatment, performance status, surgeon’s comfort with surgical approach, and ability to reconstruct the ensuing defect. ⁵ Therefore, it is especially important for physicians to set realistic expectations of cure and functional outcomes for patients to make informed care decisions.