Introduction

The issue of post-treatment surveillance of head and neck squamous cell cancer (HNSCC) after chemo-radiotherapy (CHT-RT) or RT alone has been traditionally considered as one of the most challenging diagnostic tasks even for dedicated and experienced physicians. The increasing enrollment into organ preservation protocols of patients affected by advanced-stage diseases as well as the introduction of more aggressive CHT regimens and higher-dose RT protocols have further amplified the problem of differential diagnosis between local persistence or recurrence and normal post-therapeutic sequelae such as mucositis, soft tissue edema, and fibrosis. On the other hand, the importance of an early diagnosis of CHT-RT or RT failure cannot be overemphasized, as demonstrated by the number of papers supporting the negative impact of a late diagnosis of persistence and recurrence on the chance of rescue treatment and survival [1, 2].

Significant advances in the field of imaging by refinement of conventional anatomic studies (CT and MRI) and the introduction of 18fluorodeoxyglucose positron emission tomography (18FDG-PET) and diffusion-weighted (DW)-MRI, which are based on the concept of probing the tissue microstructure by evaluation of its biologic properties (such as its glucose uptake or water content), have remarkably refined the assessment of head and neck cancer patients. Such imaging modalities have already shown promising results in post-treatment surveillance, particularly in differentiating persistence and recurrence from radionecrosis, inflammatory changes, and fibrosis [3–5].

Even when considering the endoscopic technique, standard white light (WL) evaluation has been recently implemented by the introduction of novel technologies such as narrow-band imaging (NBI) and high-definition television (HDTV) [6, 7]. NBI (Olympus Medical System Corporation, Tokyo, Japan) applies narrow-band spectrum optical filters to enhance the visualization of mucosal and submucosal microvascular patterns. This technique is based on the fact that the depth of penetration of light is dependent on its wavelength. NBI filters select the blue and green light with wavelengths of 415 and 540 nm, respectively, corresponding to the peaks of absorption of hemoglobin. These filtered wavelengths penetrate the superficial layers of mucosa, thus highlighting the capillary network, and deeper levels, enhancing the submucosal vessels, respectively. Additionally, the best image definition for both conventional WL and NBI endoscopy is achieved with the use of an HDTV camera, which gives 1,080 lines of resolution, thus allowing a signal definition that is 4.26 times better than standard definition. Even though the diagnostic value of NBI ± HDTV in the upper aero-digestive tract has been already demonstrated in different clinical scenarios ranging from preoperative evaluation to screening for synchronous and metachronous tumors, its usefulness in surveillance after CHT-RT or RT is yet to be confirmed [8–10].

The aim of the present paper was to prospectively evaluate the role of NBI ± HDTV in the follow-up of patients receiving CHT-RT or RT as part of their treatment protocol and to quantify the diagnostic improvement obtained from its use.

Patients and methods

The present study was prospectively conducted between April 2007 and April 2009 at the Department of Otorhinolaryngology, Head and Neck Surgery of the University of Brescia, Italy. During this time interval, 390 patients affected by HNSCC were evaluated by three of the authors (G.P., C.P., and D.C.) with WL and NBI ± HDTV. This analysis specifically focused on a subgroup of 59 (15%) subjects (49 males, 10 females; mean age, 62 years) previously treated by CHT-RT or RT as part of their treatment protocol, in which WL endoscopy did not show any evidence of persistence, recurrence, or metachronous lesions. These patients were evaluated during follow-up at least 5 months after treatment (range 5–24 months; mean 10). All patients received at least two evaluations, with a minimum interval between them of 6 months, and 13 (22%) were evaluated three times (for a total of 131 endoscopic evaluations).

All patients underwent transnasal flexible endoscopy using an ENF-V2 videoendoscope connected to an Evis Exera II CLV-180B light source (Olympus Medical Systems Corporation, Tokyo, Japan), after local anesthesia of the upper aero-digestive tract by 1% oxybuprocaine chlorohydrate. At present, the instrumentation cannot be integrated with HDTV for technical reasons, mainly due to miniaturization of the distal microchips (already available for transoral and transrectal flexible videoendoscopes). The nasopharynx, oropharynx, larynx, hypopharynx, and oral cavity were sequentially examined by switching from WL to NBI light using a fingertip control on the videoendoscope. All examinations were recorded and stored for subsequent re-evaluation.

We focused our attention on the early detection of persistent or recurrent disease and identification of metachronous lesions. According to literature [9, 11, 12], we considered as “suspicious” at NBI any well-demarcated brownish area with thick dark spots and/or winding vessels (Fig. 1). Furthermore, we judged the presence of an afferent hypertrophic vessel branching out into small vascular loops in the context of the lesion as “suspicious”. In case of positive NBI findings, patients were submitted to re-evaluation under microlaryngoscopy and intraoperative rigid endoscopy by WL and NBI with 0° and angled telescopes (Karl Storz, Tuttlingen, Germany) coupled to an Evis Exera II HDTV camera (Olympus Medical Systems Corporation, Tokyo, Japan).

Fig. 1
figure 1

a Examination by flexible NBI, 13 months after RT for oropharyngeal SCC followed by neck dissection. A well-demarcated dark area in the left anterior tonsillar pillar extended to the adjacent base of the tongue is visible; b closer view of the same lesion with better visualization of the typical thick brown spots

Full size image

Surgical resection was tailored taking into account pre- and intraoperative information obtained by NBI. All surgical specimens were oriented, stained with black ink at the level of one surgical margin, and submitted for histopathologic examination to a dedicated pathologist to obtain specific information on the “suspicious” areas detected by NBI.

Pre- and intraoperative videos were separately evaluated by the three examiners (G.P., C.P., and D.C.) and rated as “suspicious” or not. The examination was considered “suspicious” when at least two out of three authors concurred. All endoscopic evaluations were assessed in relation to the definitive histopathologic report to calculate the sensitivity and positive predictive rates. Although patients with a “non suspicious” NBI examination did not receive any biopsy or resection, we regarded as true negatives those patients with a persistently negative NBI follow-up evaluation to calculate the specificity, negative predictive value, and accuracy.

Results

Among 59 patients of the present cohort, 27 (46%) had been previously treated for laryngeal, 11 (19%) oropharyngeal, 11 (19%) oral cavity, 6 (10%) hypopharyngeal, 2 (3%) nasopharyngeal SCC, and 2 (3%) for metastatic lymph nodes from unknown primaries. As much as 45 (76%) of them had been previously treated by surgery followed by complementary RT (n = 37) or CHT-RT (n = 8), 8 (14%) by RT alone, and 6 (10%) by concurrent CHT-RT as an organ preservation protocol.

Flexible NBI videoendoscopy showed “suspicious” areas undetectable by routine WL videoendoscopy in 13 (22%) patients. All these lesions were confirmed as “suspicious” according to the above-mentioned criteria during intraoperative HDTV NBI, while 8 (62%) cases only were judged “suspicious” by HDTV WL. Of 13 “suspicious” lesions at HDTV NBI, 12 (92%) were histologically confirmed as carcinoma in situ (n = 6), microinvasive carcinoma (n = 2), and invasive carcinoma (n = 4). Flexible NBI, HDTV WL, and HDTV NBI findings are summarized in Table 1.

Table 1 Flexible preoperative NBI, intraoperative HDTV WL, and HDTV NBI “suspicious” examinations compared with histopathologic results
Full size table

From a practical point of view, NBI ± HDTV was useful in the early detection of incomplete response to CHT-RT or RT before planned neck dissection (n = 2), recurrences (n = 6), and metachronous tumors (n = 4). Persistences were seen less than 6 months after the end of treatment in one patient with laryngeal and one with oropharyngeal SCC. Recurrences were detected between 6 and 60 months after the end of treatment, in the oral cavity (n = 2) (Fig. 2), larynx (n = 2), oropharynx (n = 1), and hypopharynx (n = 1). Metachronous tumors were evident, with a time interval between the treatment of the index tumor and the diagnosis of the second, in the oropharynx (n = 1) and larynx (n = 3). Clinical details are reported in Table 2.

Fig. 2
figure 2

a Examination by 0° rigid HDTV WL, 47 months after right transoral partial glossectomy followed by RT. Examination shows two leucoplakias suspected to be recurrences (arrowheads); b rigid HDTV NBI showing two discrete lesions, without a typical vascular pattern (arrow heads), surrounded by mucosa with abnormal vessels (arrow) suspicious for neoplasia; c histopathologic evaluation of the excisional biopsy-confirmed HDTV NBI findings. Granuloma (arrow head) surrounded by microinvasive carcinoma is shown (arrow) (hematoxiline and eosin; original magnification, 10X)

Full size image
Table 2 Clinical data of patients “suspicious” at NBI and confirmed as true positives (n = 12)
Full size table

Overall, 12 (92%) of 13 “suspicious” lesions at NBI ± HDTV were confirmed to be true positives, while only 1 (8%) was a false positive. This latter patient had been previously treated by endoscopic supraglottic laryngectomy followed by complementary RT for a pT2N2b supraglottic tumor 2 years prior. Flexible NBI videoendoscopy showed a “suspicious” area at the level of the left vocal cord, confirmed by both intraoperative HDTV WL and HDTV NBI under microlaryngoscopy. In spite of this, excisional biopsy of the lesion revealed cheratosis without atypia. Among 118 endoscopic examinations judged as “non-suspicious” (and therefore not receiving any surgical procedure or histopathologic evaluation), we considered as true negatives 46 patients with persistently negative flexible NBI videoendoscopies during follow-up.

The sensitivity, specificity, positive, negative predictive values, and accuracy of flexible NBI, HDTV WL, and HDTV NBI are summarized in Table 3.

Table 3 The sensitivity, specificity, positive and negative predictive values, and accuracy of flexible NBI, HDTV WL, and HDTV NBI examinations for the entire cohort of patients (n = 59)
Full size table

Discussion

Zbären et al. [13] have already demonstrated that conventional WL endoscopy associated with CT scan for surveillance of laryngo-hypopharyngeal cancer after CHT-RT or RT yields a disappointing 50% accuracy in correctly staging recurrent lesions. Moreover, many authors [13, 14] have observed that in laryngeal and hypopharyngeal cancer, most recurrent tumors after RT failure appear at a more advanced stage than at initial evaluation, thus dramatically reducing the possibility of salvage surgery.

Even though great advances have been made in imaging, with the introduction of 18FDG-PET and DW-MRI, early detection of persistence and recurrence can be hampered by a temporal limit of 6–8 weeks after the end of treatment for 18FDG-PET and by the spatial resolution for DW-MRI [3–5]. Therefore, the use of an endoscopic technique like NBI that is capable of evaluating the neo-angiogenic characteristics of soft tissues “under the cover” of post-actinic edema, mucositis, and fibrosis appears to be of value in this setting. As already demonstrated by Nonaka et al. [11], NBI can distinguish the intraepithelial papillary capillary loops altered by inflammation from those observed in persistence and recurrence. In fact, in case of iatrogenic changes, vascular loops are usually diffuse, involving extensive mucosal areas, and present with ill-defined margins and a relatively low density. In contrast, “suspicious” lesions are clearly visible as discrete and well-demarcated dark areas, sometimes dispersed as multifocal spots inside inflamed mucosa (Fig. 3). In our experience, when the above-mentioned criteria are strictly followed, NBI allows detection of about 20% more persistences or recurrences than standard endoscopy. Other advantages of this endoscopic technique are that it is non-invasive, not time consuming, cost effective, and repeatable at each control visit.

Fig. 3
figure 3

Intraoperative 0° rigid HDTV NBI, 9 months after endoscopic laser surgery and postoperative RT. NBI shows a well-demarcated dark area (arrow) with the typical vascular pattern in the posterior third of the left false vocal cord (histologically confirmed to be carcinoma in situ) adjacent to inflammatory vascular changes (arrowhead)

Full size image

The utilization of NBI in the screening of the upper aero-digestive tract of patients at risk for HNSCC has already shown its value in detecting recurrences and metachronous lesions at an earlier stage than standard endoscopy [9, 10]. The present data confirm that among the 12 lesions detected by NBI, 6 (50%) were Tis, 5 (42%) T1, and only 1 case was (8%) T2. Therefore, with the exception of the latter, treated by total laryngectomy, 92% of patients in the present series had an advantage using a mini-invasive conservative surgical approach. Though the analysis of survival was not a goal of our study due to short follow-up, NBI may eventually be demonstrated to be of value in this regard.

The high sensitivity of NBI, with only one false positive case that was detected during the early phase of our learning curve (first 6 months), allowed us to significantly reduce the number of unjustified biopsies in the present cohort of patients. If one considers the possible drawbacks and risks of unneeded biopsies of the larynx and hypopharynx after CHT-RT or RT (persistent pain, non-healing ulcer, chondritis, and chondronecrosis), this certainly represents one of the most interesting advantages of this diagnostic modality [3, 13].

Conclusion

NBI ± HDTV represents a major evolution in the field of endoscopic evaluation of patients managed by multimodal treatment regimens. Its value as “functional” investigation of the soft tissues biologic characteristics in terms of neo-angiogenic growth of mucosal and submucosal vessels has been shown to be of utility for early detection of local persistences and recurrences after CHT-RT or RT. Nonetheless, anatomic as well as functional imaging with more recent technical advances such as 18FDG-PET-CT scan and DW-MRI must continue to play a complementary role, as they are the only diagnostic tools that are capable of correctly evaluating deep structures, regional lymph nodes, and distant localizations.