Introduction

In head and neck squamous cell carcinoma patients (HNSCC), post treatment surveillance for distant disease is mostly focusing on the lungs, as HNSCC distant metastasis occurs in this organ in 90% of the cases and a high rate of primary of the lungs can be expected due to field cancerization of the entire upper aerodigestive tract [1-4]. Furthermore distant metastasis isolated to other sites such as liver or bones are rare in the absence of simultaneous pulmonary malignancy [4]. The traditional method for lung screening has been chest radiography, as it is widely available, cheap and has a low radiation dose, which allows for safe repetitive screening [5,6].

Screening for distant metastasis and/or second primary of the lungs is useful as it allows for prognostication and adapted patient counselling [7] and in case of early detection may impact beneficially the prognosis of patients [8]. Early stage lung primary can be treated with curative intent [9] and some reports have claimed some benefit of metastasectomy in selected patients [8]. Overall, screening for distant metastasis has gained importance in the last few years, as aggressive locoregional treatment in HNSCC has led to better locoregional control but higher risk of distant failure [10]. This reverse failure pattern was first reported by Vikram et al. [10], who showed that only 15% of patients relapsed above the clavicle, as opposed to 70% in historical series.

Chest radiography is, however, likely to be out-dated, as previously shown by several groups [5,11]. Newer technologies now widely available such as computer tomography and positron emission tomography (PET/CT) are currently being evaluated as screening tools in HNSCC [4,5,12-14].

The objective of this study was to evaluate the current beliefs and practice among Head and Neck surgeon members of the Canadian Society of Otolaryngology using a nation-wide survey. The results of this survey were compared to latest data from the literature in the discussion.

Methods

After Ethics Review Board approval, a nationwide survey was conducted through the Canadian Society of Otolaryngology (CSO) among Otolaryngology-head and neck surgeons regarding their practices for post treatment pulmonary screening in HNSCC. A simple survey with 6 questions regarding actual practice was designed previously, reviewed, and sent to head and neck surgeons across Canada. A sample of the survey is available in Table 1.

Table 1 Structure of original questionnaire for nation-wide survey
Full size table

Results

Our CSO survey among Otolaryngologist - Head and Neck surgeons showed that 26 out 32 (81%) respondents perform routine lung screening, out of which 23 (88%) feel that chest radiography should be preferred versus low dose CT scan or PET scan for annual pulmonary screening among all asymptomatic HNSCC patients during follow up (Figure 1). In asymptomatic patients, only 2 respondents did not perform any lung screening. The response rate of the survey was 84.2% (32 out of 38) and the survey was sent only once.

Figure 1
figure 1

Preferred modality in asymptomatic patients among COS head and neck surgeons.

Full size image

For symptomatic patients, low-dose spiral CT was the preferred modality (48%), followed by PET/CT scan (14%) and sputum cytology (14%).

In high-risk asymptomatic patients (current smoker, radiation exposure, family history and advanced HNSCC), 31% of respondents performed a CXR. The same percentage performed a low dose CT, while 19% relied on PET scan. A further 19% of respondents did not perform any screening in high-risk patients.

Seventeen of 28 respondents (60%) request lung screening for 5 years, while 4 (15%) and 7 (25%) do it for 10 years and life long, respectively. Most respondents were screening their patients annually (74%), while less than 15% did it biennially or half-yearly, respectively.

CXR was believed to be in 11% “very effective” and 50% “somewhat effective” in reducing lung malignancy mortality during HNSCC follow-up for non-smoker. In former smokers, the believed efficacy was 14% and 57%, respectively. In current smoker, CXR was believed to be “very effective” and “somewhat effective” in 27% and 54%, respectively. Seventy-four percent of respondents did not recall having any patient inquiring about lung screening in the past 12 months. Most respondents (77%) had more than 10 years practice since graduation from medical school and came from the provinces of Quebec, Ontario and Alberta (11, 10, and 6, respectively).

Discussion

The main goal of this study was to assess the practice among head and neck surgeons in Canada for post treatment screening. Briefly, we showed that the preferred modality was chest radiography, done yearly in all patients for 5 years and this was believed in the majority of cases to be effective. Let us discuss each part of this statement point by point, in light of the classic Wilson and Jungner screening criteria (Table 2) [15].

Table 2 Wilson and Jungner screening criteria
Full size table

Target group

As screening is less efficacious if the expected prevalence is low according to the Bayes theorem, screening has to target particular groups in order to be cost effective and avoid high rate of false positive [16]. Previous studies have shown that patients with advanced tumor stage, tumor from the hypopharynx and smokers are at higher risk for pulmonary malignancy [1,17-19]. In human papilloma virus patients, the rate of pulmonary malignancy may be lower due to lack of field carcinogenesis and a somewhat different distant metastatic pattern [20,21], thus requiring different strategies not only in treatment but also in surveillance.

Screening accuracy and impact on mortality

Although chest radiography remains the preferred and most widely used screening modality, numerous studies showed that regular chest radiography was not efficacious at detecting pulmonary malignancy [5,6,11]. More sensitive methods are required. Computer chest tomography (CT), positron emission tomography (PET/CT) and whole body MRI (WBMRI) are actually being evaluated [4,5,12,13,22,23]. CT is now widely available, has a reported sensitivity and specificity of 73% and 86%, respectively [22] and recent development such as low dose CT may compensate for the relative high radiation dose of each examination, a factor particularly important for repetitive imaging such as in yearly screening [24]. For lung cancer, it was shown that LDCT could reduce mortality in some high-risk group such as smokers when LDCT was used as a screening tool [25]. A recent Cochrane systematic review came to the conclusion that LDCT could be beneficial in a high-risk lung cancer group, but not chest radiography or sputum cytology [26]. In HNSCC, to our knowledge, no trials are available yet.

PET/CT is deemed to detect more metastasis and earlier in their course [12,13]. This advantage however could not be translated into clinical benefit for the patients in any studies [4,13]. It must be noted that most of the studies were retrospective and that the diagnosis of distant metastasis mostly did not change the treatment, as classically only symptomatic patients will receive palliative chemotherapy but no specific intervention was performed in early diagnosis patients [27]. Classically patients with distant metastasis are treated only when symptomatic, leading some authors to argue that screening may have low clinical impact, and is unlikely to change treatment course [27]. This may be true for distant metastasis, however second lung primary may be potentially treated with curative intent and thus remains important. Furthermore, when interpreting the literature, one should be cautious not to interpret the prolonged time between diagnosis and death as survival advantage, as this may be explained by lead-time bias [6].

Although PET/CT screens the whole body and this may seem like a critical advantage, a recent study by Spector et al. [4] showed that the most common site for asymptomatic metastasis were the lungs, whereas patients who developed bone, liver or brain metastasis were typically identified by imaging initiated by symptoms. This would suggest that screening (per definition in asymptomatic patients) may be sufficient if performed in the lungs only, as metastasis to other organs are likely to become rapidly symptomatic and not “detectable” in early disease course [4]. However, in the province of Ontario, PET/CT is not approved for routine pulmonary screening and therefore would not be an option for any surgeon responding from Ontario. PET/CT scan was believed to be useful for screening by 19% of respondents (2 from Quebec; and one from Saskatchewan, Alberta, British Columbia and New Brunswick each). Baring the Ontario respondents (10 in number, 31% of total), 27% (6/22) of the respondents from provinces other than Ontario think PET/CT scan is a viable screening tool in high risk head and neck cancer patients on follow up. Finally newer studies have shown that WBMRI can be used as screening tool for HNSCC patients. WBMRI was shown in a pilot study to be accurate and not to require very long imaging times [23]. Further studies on that modality are required.

For sputum cytology, the scarce evidence available on that matter failed to support any potential use in HNSCC patients or any lung cancer case [26,28].

Timing and frequency of screening

To our knowledge, the adequate timing and frequency of screening for lung metastasis is based on empirical evidence. Most head and neck surgeons in Canada preferred yearly screening for 5 years. As distant metastasis most commonly occurs within 2 years after the diagnosis [29], one could wonder if half yearly screening in the first 2 years followed by yearly screening afterwards would be more adequate [30]. Again, although this may detect distant disease earlier in the course, it remains to be proven that this impacts survival beneficially and does not have negative impact on the health care system [31]. For screening tools with poor sensitivity such as chest radiography, a recent study showed it was ineffective when performed twice yearly [31].

Conclusion

Chest radiography remains the preferred screening modality for distant metastatic HNSCC in Canada, despite lacking sensitivity. Hence the lung screening practices of head and neck surgeons in Canada are contrary to what the evidence would suggest, and cannot be justified.