Keywords

Introduction

Head and neck squamous cell carcinomas (HNSCC) of unknown primary site (CUP) have continued to intrigue head and neck surgeons. Historically, 1.5 to 9% of head and neck cancers were considered CUP [1,2,3]. The incidence of human papillomavirus (HPV)-related oropharyngeal squamous cell carcinoma (OPSCC) has dramatically increased during the past decades. This specific group of patients is relatively young and without classical risk behaviour. HPV driven OPSCC is recognized as often presenting with tumour burden in the neck accompanied by small and occult primary tumours. This has led to an increase in CUP [4,5,6,7]. Since identification of the primary tumour has implications for treatment and prognosis, the interest in improvement of diagnostics and surgery is growing. The eighth edition of both the UICC as well as the AJCC staging manuals have incorporated the specific entity of HPV-related HNSCC with its own clinical and pathological N classification, in which a HPV-positive CUP should be classified as HPV-positive OPSCC [8, 9].

The incentives for identifying the index tumour are clear. Firstly, it is important for patients and treating physicians to have a clear view and understanding of their disease. Secondly, a true treatment target can be identified which in turn may lead to possibilities for de-intensification of treatment by minimizing toxicity in terms of xerostomia, dysphagia and also atherosclerosis of the carotid arteries. This is specifically of importance in this increasing group of young patients suffering from HPV-positive OPSCC.

History of Diagnostics in CUP

Over the previous decades, there was limited consensus in the literature with regard to diagnostic algorithms for CUP. Nevertheless, history has shown similarities and evolution in diagnostics in CUP [10,11,12]. Once an enlarged cervical lymph node has been identified as a metastasis of squamous cell carcinoma (SCC) by means of ultrasound guided fine needle aspiration (FNA) or core needle biopsy (CNB), a thorough history and physical examination combined with office-based endoscopy is a key step in initial diagnostics. CNB should be considered a reliable alternative in case of non-diagnostic FNA. The introduction of new CNB techniques may permit for more tissue yield in one needle pass and may further reduce the risk of seeding [13, 14]. When no abnormalities are found, a diagnostic contrast enhanced computer tomography (CT) or magnetic resonance imaging (MRI) of the head and neck is performed, often combined with positron emission tomography (PET). CT and MRI are not only useful in identifying the possible index tumour but can also delineate non-palpable nodes (e.g., retropharyngeal) thus pointing in a certain direction in search for the primary tumour [14, 15]. Endoscopy under general anaesthesia (EUA) with appropriate biopsies and palatine tonsillectomy (PTE) is consecutively performed. Historically, despite this proper work up, the identification rate has not exceeded 50% [16,17,18].

Currently, it is general practice to at least perform p16 immunohistochemistry (IHC) after FNA or CNB in case of proven cervical metastasis of SCC. When feasible, true HPV-positive disease needs to be confirmed by polymerase chain reaction (PCR) or in situ hybridization (ISH) testing to isolate HPV DNA. The surrogate marker p16 has led to false positive results leading to misjudgment of the potential primary tumour site. Especially in cytology specimens current literature advises to perform direct HPV testing [19,20,21,22,23]. Epstein Barr virus (EBV) testing is optional in case of suspicion of (mostly non-keratinizing) nasopharyngeal carcinoma (NPC) [24, 25]. Non-targeted biopsies are currently considered obsolete due to their low yield [26]. PTE is considered superior to biopsy of a clinically non-suspicious tonsil.

Recent and Current Improvements

With regard to morphologic and functional imaging, specifically MRI has improved regarding resolution as well as with optimization of diffusion weighted (DW) MRI. Adding DW-MRI to 18F−fluorodeoxyglucose (FDG) PET-CT does not seem to improve the sensitivity and specificity compared to PET-CT alone in detecting CUP. DW-MRI might be an alternative to PET-CT in detection of CUP in case PET-CT is unavailable. The costs of DW-MRI are lower. PET-CT on the other hand has the advantage of being a whole-body examination with synchronous screening for distant metastasis [27]. Besides PET-CT, PET-MRI seems to be a meaningful adjunct [28]. The usefulness of PET either combined with CT or MRI for identification of occult tumours in the oropharynx is somewhat limited due to the physiological uptake in the mucosa-associated lymphoid tissue at this site. The Achilles heel of PET-CT remains the relatively low specificity [27, 29]. Nevertheless, reported identification rates of occult tumours with 18F- FDG PET-CT vary from 24.5 to 40.5% [30,31,32,34]. In a recent study by Stadler et al. PET-CT/MRI was superior in staging of the neck compared to ultrasonography. Discordance varied between 20 and 65% with the majority of cases being upstaged after PET-CT/MRI [35].

Image guided surgery (IGS) has become widely adopted in general. Narrow band imaging (NBI) has proven its benefit in head and neck cancer management and in CUP specifically. It can be used in office-based endoscopy. Its concept is based on the use of blue and green light with different wavelengths, optimizing visualization of changes in mucosal microvascular patterns suggestive for dysplasia and malignancy [36]. There have been reports on a 35% added identification percentage after a negative PET-CT and MRI [37]. Others have described high negative predictive values and high sensitivity [38, 39]. The use of high definition (HD) and ultra HD cameras with 3D and 4K technology has led to superior visualization of surface mucosa. Performing endoscopy assisted office-based biopsies is becoming more and more routinely applied due to this optimization of visualization. However, when no target for biopsy is found, EUA is evidently superior due to easier access and the possibility for meticulous palpation of the mucosa [14].

Algorithms for CUP over the past 5 years have shown increasing uniformity with a prominent role for tongue base mucosectomy (TBM), also called lingual tonsillectomy (LT) in HPV-positive CUP [1, 40,41,42]. Until recently, the role of NBI had not been specifically described in guidelines for CUP. The guideline by the American Society of Clinical Oncology (ASCO) was actually the first to recommend and incorporate NBI in the guideline for CUP in 2020 [40].

The main reason for debate remains what to do in case of a true CUP after full work up. Should treatment of the neck suffice either by neck dissection (ND) or RT (in case of single node without extranodal extension; ENE) or by chemoradiation for multiple nodes or in case of ENE? Should the oropharynx be irradiated in case of HPV-positive disease and to what extent?

Transoral Surgery—The Role of TLM and TORS

To improve the identification ratio in CUP, TBM has gained popularity in the diagnostic work up. Different guidelines generally advise to perform this in case of negative work up and negative PTE for (HPV) positive nodes in levels I, II, III and upper V [40,41,42]. Farooq et al. have reported on an identification ratio by TBM of the index tumour of 78% in case of initial negative radiology, EUA and PTE [43].

This can either be done by transoral laser microsurgery (TLM) or by means of transoral robotic surgery (TORS). The use of simple tonsillectomy instruments in performing TBM has also been described by Davies-Husband [44]. The limited line of sight and oropharyngeal accessibility of TLM makes it less useful in a subset of patients. TORS allows for en bloc resections and a theoretically better margin assessment than TLM [45, 46]. The availability of TLM in most head and neck referral centers however allows for easy access to this method. In a recent study in patients with demonstrated oropharyngeal cancer by Parimbelli et al. the cost- effectiveness of TLM vs. TORS was analyzed. Their results suggest lower costs for TLM with the footnote that this depends on the number of re-interventions and the need for adjuvant treatment [47]. The most advocated method to perform TBM is through TORS, executed with the da Vinci surgical system by Intuitive Surgical© (Sunnyvale, California; US). A surgical robotic system is becoming more readily available throughout centers since other surgical specialties have increasing indications for robotic surgery such as urological surgery and gastro- intestinal surgery. Due to the use of the robotic arms and angled high-definition endoscopes as well as wristed instruments, the oropharynx is readily accessible in the majority of patients. Currently, the da Vinci Si and Xi surgical systems are the most widely used. The single port (SP) system might be beneficial in TORS. The first reports on its use in TORS are promising in terms of safety and outcome compared to the conventional multi- port systems used. Specifically for transoral surgery of the distal Upper Aero-Digestive Tract (UADT), the SP system seems superior in terms of accessibility [48,49,50].

TBM has earned its place in different international guidelines regarding CUP such as those from ASCO as well as the National Comprehensive Cancer Network (NCCN) [38, 40]. Several European authors have suggested modified CUP algorithms, which are comparable with overall minor differences between them [1, 51].

Where Does TBM Come In?

Although there is sufficient proof for the added value of TBM, the question remains as to when and how to perform TBM [16, 52, 53]. It is generally agreed that TBM comes into play in case of a negative office- based endoscopy (+NBI when available), a negative PET-CT (and MRI) and subsequent negative EUA and PTE. Several previous systematic reviews have shown that the pick- up rate of occult tumours is relatively high in the tongue base as compared to the palatine tonsils (Table 12.1) [16, 43, 54, 55]. The pooled identification rate of the index tumour is 72% (range 25–100%) as reported in the review by van Weert et al. (Table 12.2) [16, 52, 53, 56,57,58,59,60,61,62,63,64,65]. There is no consensus as to whether to perform staged (second procedure after negative EUA and PTE) TBM or not. Those in favor of a staged procedure would argue that patients have less pain and less feeding tube dependence and can avoid TBM in case of a positive PTE. Another reason might be a swifter recovery as not to further postpone possible radiotherapy (RT). Post-surgical pharyngeal stenoses have been described after a simultaneous procedure. A staged procedure seems cost- effective as reported by Byrd et al. [53].

Table 12.1 Occult tumour identification rates for TORS/ TLM. Note the relatively high pick- up rate in the base of tongue. TORS; transoral robotic surgery, TLM; transoral laser microsurgery, TBM; tongue base mucosectomy, PP; pooled proportion
Full size table
Table 12.2 Identification rate for occult tumours with TORS. Range 25−100%
Full size table

Although positive TBM procedures have been described in HPV-negative disease, the wide majority are HPV-positive [16, 66]. This is due to the high prevalence of CUP in the HPV- positive group as well as the strong correlation between HPV and OPSCC. There is no strict consensus regarding the usefulness of TBM in the HPV-negative population although recent literature supports the omission of TBM in HPV-negative disease [67].

How to Perform TBM?

The technique of TBM has been previously described [58, 61, 64]. Starting off with a midline incision from the level of the circumvallate papillae towards the vallecula, the incision is then carried through laterally at the level of the anterior base of tongue (BOT). Once the glossopharyngeal sulcus is reached, the incision is extended towards the vallecula. The tongue musculature marks the depth limit as to only remove the lymphoid tissue. After visualization of the pharyngoepiglottic folds, the first half of the BOT is removed (Fig. 12.1). There are modifications in this technique where for example some authors do not use the midline incision. The general concept however remains the same.

Fig. 12.1
Two photographs of tongue base mucosectomy.

Intraoperative images of a left-sided tongue base mucosectomy (TBM) performed with the da Vinci Si system. The first image shows the initial midline incision with the monopolar cautery. In the second image, the specimen is retracted backward with the Maryland dissector for the final incision in the vallecula

Full size image

How to Analyze the Resection Specimen?

After the surgeon has adequately orientated the specimen (preferably with needles on cork with the mucosa facing the cork in order to prevent desiccation [64]), it is sent to the pathology laboratory with a proper description of the clinical situation. In theory, ideally the pathologist would perform step serial sectioning (SSS). This entails a section every 0.5 mm through the entire specimen and is known from other head and neck procedures such as the sentinel lymph node biopsy. This is a very time-consuming procedure, which should harbor clear benefit in comparison to conventional sectioning. Recent preliminary data from the Royal Marsden in London through the clinical trial “Evaluation of Tongue Base MucOsectomy & Step sErial Sectioning (MOSES)” has shown that the actual added value of SSS seems relatively small. These data were presented at the ICHNO-ECHNO meeting in March 2022 in Brussels [68]. This poses the question whether the ends justify the means with regard to SSS in TORS for CUP. IHC for p16 on the specimen however may play an important role in identifying occult tumours of the tongue base (Fig. 12.2) [16].

Fig. 12.2
Two microscopic images of the cystic squamous cell carcinoma.

Small (2.2 mm) partly cystic squamous cell carcinoma detected in the crypts of the lymphoid tissue in the base of tongue in a patient with multiple unilateral metastases in the neck. (left hematoxylin and eosin, right p16 immunostaining) [16]

Full size image

Extent of Resection

Historically, there have been reports on occult tumours being found in the contralateral oropharynx with rates of up to 10% [69]. Geltzeiler et al. reported on an 80 versus 68% identification rate for bilateral vs. unilateral TBM, respectively [52]. Nevertheless, there has never been a clear consensus on this topic within the previous guidelines; these do not clearly state as to whether to perform uni-or bilateral procedures in which scenario. The ASCO guideline from 2020 has made several recommendations on this topic [40].

The wide majority of patients with CUP present with unilateral neck nodes. This of course implies a high chance of an ipsilateral oropharyngeal occult primary, specifically in case of level II nodes, which are most prevalent in CUP. The ASCO recommendation is to perform unilateral PTE in case there is no clinical suspicion of contralateral disease. In case a frozen section is negative, the surgeon may proceed to perform an ipsilateral TBM. In the event a frozen section is not done, the procedure should be performed in a staged fashion. Moreover, frozen sections are notably less reliable than definitive histopathology and require extra surgical time. In case a contralateral PTE is performed and a small focus of SCC is found, it should be assumed that this tumor is metachronous to the yet undiscovered ipsilateral primary actually causing the nodal tumour burden [69]. Paleri has recently argued to use the term “MALTectomy” (MALT: Mucosa Associated Lymphoid Tissue) -which means both PTE and TBM- and to perform this procedure unilaterally in case of a single HPV-positive neck node without signs of ENE. The identification rate in unilateral MALTectomy in the series reported by Paleri was as high as 85%. Specifically, identification of occult disease in the contralateral palatine tonsil was seldom seen. These findings are supported by the review of Farooq, in which 97% of the primary tumors were found ipsilaterally in 556 cases. The primary tumour was found contralaterally in 2% of cases and synchronous bilateral tumours were found in 1% [43].

The scenario of bilateral neck disease in CUP is far less prevalent and presumably caused by a BOT tumour (levels II-III) being considered a “midline structure”. According to this rationale, it is recommended to perform ipsilateral TBM on the side of the largest nodal burden. This can then be extended to a bilateral TBM in case the initial hemi- TBM fails to identify the tumour. There is no true incentive to perform PTE in bilateral neck nodes given the rarity of bilateral neck metastases in SCC of the palatine tonsils [58, 70].

The Case of True CUP

A “true” case of CUP would be best defined as an ultimately unidentified primary tumour after full diagnostic work up including removal of mucosa associated lymphoid tissue. Several theories have been proposed addressing the issue of the non- appearing index tumour even after long term follow up following ND only without radiation therapy. One possibility could be that these tumours regress spontaneously, possibly due to immunological factors, as is known for other tumour types [71, 72]. Another scenario would be the speculation by Califano et al. which was also included in the review by Civantos et al.: in 56% of cases with random biopsies from possible primary tumour sites they found genetic alterations identical to those in the neck nodes [14, 73]. These were described by Civantos as genetically malignant but phenotypically benign. In this case, these genetic alterations would be able to cause metastatic disease to the neck without inducing macroscopic disease at the primary site [14]. This intriguing issue will remain a topic of discussion in the years to come.

Is TBM Really Indispensable?

Identification of a primary lesion leads to appropriate staging and understanding of the disease. This is of importance for both the patient and the treating physician. Striving for single modality treatment whenever possible is important. In case the occult primary has not only been found but also adequately excised, the oropharynx can be left out of the field of radiation. This in turn will have a beneficial effect on costs, specifically in case the neck needs no radiation after ND. In case of an insufficient margin, re-excision should be contemplated. This will often be an option given the small volume of these occult tumours. Even if this would not be possible, the field of radiation may be minimized due to the successful localization of the primary tumour.

Previous studies have reported on superior outcome in case of discovering the primary tumour in the pre- HPV era [74, 75]. Karni et al. confirmed this finding with their report on TLM in CUP with a high identification rate in the oropharynx of up to 95%. HPV status was not analyzed in this study [76].

In 2014, Davis et al. reported on a significantly higher identification rate in case of HPV-positive disease which in turn was positively correlated with overall- and disease specific survival (Figs. 12.3 and 12.4). Of course, one can never be sure about the contributing factor of HPV-positivity on these outcomes regardless of identification of the occult tumour. Furthermore, this study also emphasizes that TBM in HPV-negative disease is debatable [66].

Fig. 12.3
A line graph plots the cumulative survival with respect to the overall survival. It involves clinical follow-up in years.

Overall survival relative to identified and non- identified primary tumours [66]. Reprinted with permission ©

Full size image
Fig. 12.4
A line graph plots the cumulative survival over a period of time with respect to disease-free survival. The line represents the primary not found, primary found, primary not found censored, and primary found censored.

Disease- free survival relative to identified and non- identified primary tumours [66]. Reprinted with permission ©

Full size image

Definitive Treatment

In case the primary tumour is identified after thorough work up, the tumour should be staged and treated accordingly. The primary objective should be to avoid using three treatment modalities when possible to reduce treatment related long-term toxicity. Specifically in the current era of increasing incidence of HPV-associated OPSCC, there is a clear incentive for de-intensification of treatment in this mainly young patient population with their whole lives ahead of them. Early stage OPSCCs are treated either by primary radiation or by primary surgery.

The ideal indication for surgery alone should be a clinically single unilateral node without signs of ENE and a potentially resectable tumour that allows for proper margins, however these indications where the neck is concerned might be extended in the future. What a proper margin is in TORS for OPSCC has not been clearly defined. There have been recent reports supporting the theory that margins less than 1 mm (herein defined as close) might justify active surveillance at least in HPV-positive OPSCC. In this report by Holcomb et al., there was no significant correlation with close vs. clear margins (defined as >1 mm) relative to local control, disease-free and overall survival. Outcome did not improve with margins of 2 to 3 mm [77].

In case no tumour has been found in the CUP work up, ND alone might suffice with meticulous periodic office-based endoscopy surveillance. One should bear in mind that the eventual pathology report might up-stage the pN classification in up to 30% of cases when identifying additional positive nodes or ENE, which would necessitate postoperative (chemo)RT to the neck [17, 78]. It is important to consider this and discuss it with the patient in advance. A recent report by Grewal et al. confirmed the safe possibility of pharyngeal sparing RT (PSRT) as opposed to pharyngeal RT (PRT) in case of a pT0 TORS work up. PSRT is defined as RT to the neck at risk (with or without upfront ND) with omission of RT to the pharynx. PSRT did not compromise outcome and significantly reduced pharyngeal toxicity [79]. Other reports have described comparable outcomes in terms of locoregional control and survival in case of unilateral RT to the neck without irradiating the pharyngeal axis as opposed to the classical bilateral neck irradiation including the pharyngeal mucosa [80, 81]. Tiong et al. reported on good results on unilateral treatment for HPV-positive OPSCCs specifically, with no contralateral neck failure or primary emergence [82].

Since the entity of CUP within the spectrum of head and neck cancer remains relatively rare, most studies do not encompass large cohorts and most of the knowledge currently shared is based on systematic reviews, which have been published in quite a large amount over the last years. These reveal that treatment paradigms still differ internationally and even nationally [16, 40].

With the input of current literature on CUP, different scenarios can be suggested based on the outcome of TBM, HPV status, N-status and the presence or absence of ENE after negative EUA and PTE. A proposed algorithm based on current insights is shown in Fig. 12.5.

Fig. 12.5
An algorithm depicts the F N A forward slash C N B proven S C C neck nodes. It suggests the different scenarios based on the outcome of T B M, H P V status, N-status, and the presence or absence of E N E after negative E U A and P T E.

Proposed algorithm for CUP and the role of TBM (tongue base mucosectomy). FNA; fine needle aspiration, CNB; core needle biopsy, IHC; immunohistochemistry, PCR; polymerase chain reaction, ISH; in situ hybridization, NBI; narrow band imaging, EUA; endoscopy under general anaesthesia, PTE; palatine tonsillectomy, CRT; chemoradiation, ENE; extranodal extension, ND; neck dissection, FU; follow up, PSRT; pharyngeal sparing radiotherapy. *; unilateral TBM suggested to be initially performed in selected cases (see ASCO guideline) [40]

Full size image

Conclusion

Although CUP in head and neck cancer is in general rare, the rising incidence of HPV-positive OPSCC has drawn new attention to the topic. Evolutions in radiology and endoscopy as well as the use of transoral surgical techniques have evidently contributed to a higher identification rate of these occult tumours. The historical discovery rate in the “naked eye” era with EUA and PTE with or without blind biopsies was approximately 50% at best and has now increased to approximately 80%. TBM has been adopted in every current guideline or algorithm, mainly in HPV-positive neck nodes in levels II, III and Va [1, 40, 83]. TBM has not only proven to be useful in discovering occult tumours but may also achieve an adequate resection of the index tumour, thus making PSRT possible. In the optimal case of a unilateral single neck node without ENE, TBM combined with ND can suffice as a single modality treatment; in case the primary tumour is excised or remains undetected (pT0), a periodic office based endoscopy of the pharynx seems a reasonable policy. This is even more reliable due to the HD features and possibilities of using NBI. Discovery of the primary lesion improves survival and allows for targeted and de-intensification of treatment. The use of TBM (and PTE) in HPV-negative CUP is questionable since the detection rate in this category has proven to be very low [16, 66, 84]. Treatment strategies still differ among centers with regard to for example upfront ND or primary RT. The question is whether this is a true problem, as long as outcome in terms of survival and quality of life are comparable. Questions that have been raised recently mainly concern the extent of TBM and the necessity of performing bilateral PTE. Based on the data currently at hand, there seems to be an incentive for initial unilateral MALTectomy. Further research is mandatory to confirm this.