Utility of Surveillance Imaging in Patients with Head and Neck Cancer Treated with Definitive Radiotherapy

Abstract

Purpose

To assess the efficacy of surveillance imaging in patients with head and neck cancer (HNC) treated definitively with radiotherapy.

Patients and Methods

Eligible patients included those with a demonstrable disease free interval (≥ 1 follow up imaging without evidence of disease and a subsequent visit/ imaging) treated for HNC during 2000–2010.

Results

1508 patients were included. Median overall survival was 99 months with a median imaging follow up period of 59 months. 190 (12.6%) patients had disease recurrence – 107 locoregional and 83 distant. 119 (62.6%) of the relapsed group were symptomatic and/or had an adverse clinical finding associated with the recurrence. 80% of locoregional relapses presented with a clinical finding, while 60% of distant relapses were imaging-detected in asymptomatic patients. Despite earlier recurrence detection via imaging, those who were in the clinical-detected recurrence group were significantly more likely to receive salvage therapy than those who were imaging-detected (OR = 0.35). There was no difference in overall survival between those with relapses detected clinically or with imaging alone. 70% of relapses occurred within the first 2 years. In those who relapsed after 2 years, the median time to relapse was 51 months. After 2 years, the average number of imaging per patient for detecting a salvageable recurrence for imaging-detected group was 1539.

Conclusions

Surveillance imaging in asymptomatic patients treated definitively with radiotherapy without clinically suspicious findings beyond 2 years has a low yield and high cost. Physicians ordering these studies must use judicious consideration and discretion.

Introduction

Head and neck squamous cell carcinoma (HNSCC) has an incidence of approximately 50 000 annual cases in the United States, with an annual mortality estimated at 10 000 persons. ¹ Radiation therapy plays a major role in the curative treatment of a majority of head and neck cancer cases – either in the definitive or post-operative setting. In the post-treatment setting, patients are evaluated for evidence of disease recurrence and late effects of treatment. Routine clinical surveillance with imaging is frequently obtained at fixed intervals as determined by institutional or departmental policy. The goals of surveillance are to evaluate and manage late effects of treatment and to detect potentially asymptomatic salvageable recurrences or second primary tumors. More recently, studies have questioned the need for routine imaging of asymptomatic patients who achieved complete clinical and radiological response after radiotherapy. ^2–6 The objective of this study were to assess the efficacy of imaging, in the modern era, compared to clinical evaluation in detecting early salvageable recurrences.

Methods

Eligible cases included those treated with definitive radiotherapy with or without chemotherapy at The University of Texas MD Anderson Cancer Center, USA between year 2000 and 2010. All cases had a demonstrable disease free interval with at least 2 follow up visits after treatment and ≥ 1 follow up imaging without evidence of disease. Age at diagnosis, tumor site and stage, use of neo-adjuvant chemotherapy, radiation dose and fractionation, mode of detection of recurrence, salvage therapy, number and modality of routine surveillance imaging were recorded. Position emission tomography (PET) imaging was commercially introduced during this period, therefore the utility of PET imaging in the surveillance period was recorded. Human papillomavirus (HPV) and p16 status were not routinely reported prior to 2010 and therefore, were not collected. Deaths from disease recurrence or from other causes were also recorded.

Recurrences were classified as either clinical-detected or imaging-detected. Clinical detection was defined as recurrences which were symptomatic and/ or visible on clinical assessment (which included history-taking, complete physical head and neck examination, and mirror or fiberoptic nasoendoscopy examination). All clinical assessment were performed and documented by an attending physician. Recurrences that were detected on imaging alone in an otherwise asymptomatic patient with no gross adverse clinical findings were documented as imaging-detected. Salvageable recurrences were defined as recurrences that were subsequently treated with curative intent, with surgery and/ or radiotherapy.

This retrospective study was approved by our hospital’s Institutional Review Board (IRB).

Statistical analysis

Overall survival was calculated from the date of completion of radiotherapy to the date of death from any cause, or date last known alive. Relapse-free survival or time to relapse was calculated from the date of completion of radiotherapy to the date of disease recurrence. Overall survival from relapse was determined from the date of relapse to date of death from any cause, or date last known alive. Survival outcomes were estimated using the Kaplan-Meier method and groups (relapse versus no relapse, and imaging- versus clinical-detected recurrences) were compared using the log-rank test. P-value ≤ 0.05 was considered statistically significant. Weibull distribution competing cause analysis for probability of recurrence over time was performed.

Average number of imaging for detecting a salvageable recurrence was calculated based on total number of surveillance imaging of the group divided by the number of patients with salvageable recurrence and those who had a successful salvage outcome, defined as being alive or dead not secondary to disease recurrence.

All statistical analyses were performed using the JMP package (v 12.1.0, SAS Institute Inc.).

Results

Baseline characteristics

From 2000 to 2010, 1508 patients were identified as eligible for analysis. The median age at diagnosis was 55.8 years (range: 17–87 years).

The majority of the cohort had oropharyngeal cancer (82%) and had nodal involvement. Neo-adjuvant chemotherapy was utilized in 29.7% of cases. 635 (42%) patients had PET imaging at some point during surveillance. Table 1 summarizes the patient, tumor, and treatment characteristics for the cohort.

Table 1.

Patient and tumor characteristics

	N (n = 1508)	%
Age at diagnosis
Median	55 years
Range	17 – 87 years
Primary tumor site
Oropharynx	1240	82.2
Nasopharynx	82	5.4
Larynx	62	4.1
Hypopharynx	38	2.5
Oral Cavity	6	0.4
Sinonasal	12	0.8
Unknown primary	68	4.5
Tumor (T) stage
T0	68	4.5
T1	367	24.3
T2	498	33
T3	300	19.9
T4	223	14.8
Tx	52	3.5
Nodal (N) stage (all except nasopharynx)
N0	156	10.3
N1	189	12.5
N2a	135	9
N2b	585	38.8
N2c	223	14.8
N3	97	6.4
Nx	41	2.7
Nodal (N) stage (Nasopharynx)
N0	16	1.1
N1	22	1.5
N2	36	2.4
N3	8	0.5
Smoking status at diagnosis
Current	354	23.5
Previous (≥ 10 pack years)	404	26.8
Previous (< 10 pack years)	206	13.7
Never	544	36.1
Concurrent chemotherapy
Yes	1003	66.5
No	505	33.5
Induction chemotherapy
Yes	448	29.7
No	1060	70.3
Radiotherapy dose
Median	6996 cGy
Range	5710 – 7600 cGy
Number of fractions
Median	33
Range	28 – 64

Patterns of failure and method of detection

The median imaging follow-up period was 59 months (range: 1 – 196 months). Patients had a mean of 14 routine surveillance imaging performed.

One hundred ninety patients (12.6%) had disease recurrence with 107 locoregional, 12 local or regional with distant disease, and 83 distant relapses as first presentation (Table 2). One hundred nineteen patients (62.6%) had recurrences that were clinical-detected, while 71 patients (37.3%) had recurrences that were imaging-detected. Clinical-detected recurrences were found later than imaging-detected recurrences with a median time to recurrence of 16 versus 11 months (p < 0.0001), respectively. Patients who were current smokers or previous smokers with ≥10 smoking pack-years developed disease recurrence significantly earlier than never smokers or those who had < 10 smoking pack- years (median time to recurrence = 11.5 versus 16 months, p = 0.0241) (Supplemental figure 1).

Table 2.

Sites of initial disease recurrence and number of patients who had salvage therapy

Site of recurrence	N (total)	%	N with recurrence		N received salvage therapy (n=73)
			Clinical-detected	Imaging-detected	Clinical-detected	Imaging-detected
Local	48	25.3	45	3	30	1
Regional	40	21.1	29	11	21	8
Local & Regional	7	3.7	5	2	2	1
Local & Distant	6	3.2	5	1	0	0
Regional & Distant	6	3.2	2	4	0	1
Distant	83	43.7	33	50	3	6
Total	190				56	17

One hundred thirty-three recurrent events (70%) occurred within the first 2 years after completion of definitive radiotherapy (Figure 1). Of those 133, 75 were clinical-detected and 58 were imaging-detected. Fifty-seven recurrences (30%) developed after 2 years post-treatment; 44 were clinical-detected and 13 were imaging-detected (2 had locoregional and 11 had distance recurrences). The overall median time to relapse was 12.5 months (range 1 – 160 months), while the median time to relapse for those relapses found after 2 years was 51 months (n = 1413 patients at risk after 2 years). Weibull distribution competing cause analysis showed that the probability of detecting any recurrences with imaging alone in an asymptomatic patient is very low (Supplemental figure 2) and imaging is more likely to detect asymptomatic distant metastatic disease rather than locoregional disease (Supplemental figure 3).

Figure 1.

Failure plot depicting the number of disease recurrences over time, with the majority of the recurrences occurring within the first 24 months.

Outcomes and Salvage Therapy

The median follow up time for all patients was 99 months (range: 6 – 199 months). The actuarial 2-, 5-, 10-year overall survival rates were 93.7%, 86.2% and 71.9%.

Seventy-three patients (38%) received salvage therapy (Table 3). Of these, 62 were clinical-detected and 17 imaging-detected. Forty-five and 28 patients with recurrences occurring ≤ 2 years and > 2 years post treatment, respectively, underwent salvage therapy. Despite earlier recurrence detection via imaging, those who had recurrence detected clinically were significantly more likely to receive salvage therapy than those who were imaging-detected (OR = 0.35, p = 0.0015). Distant disease recurrences were more common in the imaging-detected group. There was no difference in overall survival and survival from salvage treatments between those with disease recurrence detected clinically (n=56) or via imaging (n=17) (Supplemental figure 4).

Table 3.

Number of recurrences and salvage treatments detected within and after 24 months

Sites	Within 24 months			After 24 months
	N	Attempted salvage	Successful salvage	N	Attempted salvage	Successful salvage
Local	26			22
	- 23 clinical	16	10	- 22 clinical	14	9
	- 3 imaging	1	0	- 0 imaging	0	0
Regional	31			9
	- 21 clinical	15	11	- 8 clinical	6	4
	- 10 imaging	7	5	- 1 imaging	1	1
Local & Regional	6			1
	- 5 clinical	2	1	- 1 imaging	0	0
	- 1 imaging	1	0
Local & Distant	5			1
	- 4 clinical	0	0	- 1 clinical	0	0
	- 1 imaging	0	0
Regional & Distant	6			0
	- 2 clinical	0	0
	- 4 imaging	1	1
Distant	59			24
	- 20 clinical	1	0	- 13 clinical	2	2
	- 39 imaging	1	1	- 11 imaging	5	5

Average number of surveillance imaging

Table 4 sums up the total number of imaging procedures performed for each imaging modailty. In the first 2 years post definitive radiotherapy, 45 patients had a salvageable recurrence with 34 within the clinical-detected group and 11 within the imaging-detected group. The average number of imaging procedures performed to detect an asymptomatic salvageable recurrence within 2 years of definitive treatment was 1108. The average number of locoregional imaging (no chest/ abdomen /pelvis imaging) required to detect an asymptomatic salvageable locoregional recurrence was 784.

Table 4.

Sum of surveillance imaging done for the cohort (1508 patients). CT – computed tomography; MRI – Magnetic resonance imaging; FDG – fluorodeoxyglocose; PET – positron emission tomography.

Imaging modality	N (first 24 months)	N (after 24 months)	Total (%)
Chest X ray	4504	4924	9428 (42.8)
CT soft tissue neck with contrast	5658	3290	8948 (40.6)
CT chest with contrast	602	551	1153 (5.2%)
CT abdomen/ pelvis with contrast	27	20	47 (0.2%)
MRI head/ orbit	527	443	970 (4.4%)
Ultrasound neck	107	287	394 (1.8%)
18-FDG PET/CT	766	314	1080 (4.9%)
Total	12191	9829	22020

After 2 years of disease-free period, 28 patients had a salvageable recurrence – 22 within the clinical-detected group and 6 within the imaging-detected group. Overall, in this cohort, the average number of scans required to detect a salvageable recurrence in an asymptomatic patient after 2 years was 1539. The average number of locoregional imaging required was 4334. Given that the majority of recurrences occurred within the first 2 years post-treatment with 47% detected via imaging alone, it is reasonable to perform surveillance imaging within this follow-up time period.

Discussion

The primary aim of surveillance imaging is to detect a potentially asymptomatic salvageable recurrence or second primary tumor, allowing early salvage treatment; presumably this will improve subsequent clinical outcomes. In this study of a large cohort of patients with head and neck cancer treated definitively with radiotherapy with or without chemotherapy, we found that, firstly, 2 out of 3 recurrences presented with a clinical symptom or an adverse clinical finding. Secondly, the majority of disease recurrences occurred within the first 2 years of surveillance. The yield of detecting a salvageable recurrence with routine imaging after 2 years in an asymptomatic patient with no adverse clinical finding is extremely low (3% in this cohort). Although surveillance imaging detected recurrent disease earlier, there was no difference in the salvage or survival rates when compared to those with clinical-detected disease. Therefore, our study highlights that thorough history and clinical assessment is of upmost importance in surveillance, particularly after 2 years post-treatment, while imaging may be reserved for those with unfavorable clinical assessment.

Currently there is no clear consensus with regards to the optimal timing and modality of imaging in the surveillance period. There are no randomized control trials comparing a pre-defined surveillance imaging strategy with no follow-up imaging at all. Although the current National Comprehensive Cancer Network (NCCN) guidelines ⁷ recommend that a history and physical examination be performed at every follow-up visit, reserving additional surveillance imaging to be repeated or done routinely annually if clinically indicated, many physicians endorsed routine surveillance imaging at each follow-up visit.^8,9 Physicians may not necessarily adhere to recommended guidelines for one or more of the following reasons: lack of agreement with the guideline, lack of self-efficacy (e.g. physician believes that his/her examination skill is inadequate), inertia of previous practice (habit), external barriers (e.g. patient’s preference), and environmental factors (e.g. perceived increased in malpractice liability).⁹ With the annual direct cost of cancer care projected to rise to more than $170 billion in year 2020 ¹⁰, judicious allocation and use of resources need to be considered.¹¹

In our cohort of patients, the majority with disease recurrence presented with symptoms or had an adverse clinical examination finding. This is consistent with other historical studies in both head and neck cancer ^12–14 and other tumor sites, such as breast cancer ^15–18 and lymphoma. ¹⁹Studies in other tumor sites, such as breast ¹⁵ and lymphoma ^19,20, have shown a minimalist approach to routine surveillance imaging to be a cost-effective approach without compromising patient survival outcomes. Consistent with our data, studies in head and neck cancer have previously shown that patient symptoms and/or adverse clinical examination findings accounted for the majority of presentation of salvageable recurrences.^12,14 Our study also showed that although imaging did detect recurrences earlier, those who had a clinical-detected recurrence were more likely to receive salvage therapy. Those with imaging-detected recurrences tend to have distant disease, thereby limiting the option of salvage therapy, while those with local and/or regional recurrences were more likely to present with an adverse clinical finding.

While there has been no study indicating that frequent surveillance imaging translated to improved patient survival, surveillance imaging should not be entirely dismissed. Kissun et al ¹² demonstrated that while the majority of their cohort developed symptoms prior to the formal diagnosis of a recurrence, less than half of those who were symptomatic sought early medical assistance. It remained questionable whether the lag time between symptom development and formal diagnosis of a recurrence contributed to poorer outcome of the patients. Given that our study finding, consistent with other studies ^14,21, has demonstrated that the majority of relapses occur within the first 2 years post-treatment, it may be reasonable to consider regular surveillance imaging during this period and transition to clinical follow-up alone after 2 years post-treatment, with surveillance imaging reserved for certain high-risk patients such as smokers.

As the majority of imaging-detected asymptomatic recurrences were distant metastases, the prognosis for this group of patients is poor. Some may argue that early detection of incurable disease may potentially add distress and anxiety to patients and their families as those with small volume disease burden and/or slow disease progression may be observed. In the era of increasing HPV-related oropharyngeal cancer and development of novel immunomodulating agents and radiotherapy techniques such as stereotactic radiotherapy, asymptomatic patients with small disease burden may now be offered systemic immunotherapy agents and/or stereotactic irradiation. These treatments may be effective in achieving disease control in those with limited disease and are generally well-tolerated.

This study has limitations. Firstly, this study comes with all the caveats of a single institution retrospective cohort study. As most patients had imaging performed prior to clinical evaluation, and subsequently reported symptoms to the physician and/or had gross clinical findings that were recorded in medical records, there may be an overestimation of recurrence classified as clinical-detected. We have taken care to only group these patients as ‘clinical-detected’ if the medical records stated that the patient had obvious clinical symptom prior and/or had obvious clinical examination findings. Patients with recurrence whose clinical records described ‘subtle’, ‘vague’ or ‘mild irregularity’ findings were grouped as ‘imaging-detected’ to minimize bias. Secondly, we reported a heterogenous group of patients and treatments. Patients were of different head and neck cancer primary sites and of all stages. Although all patients received definitive radiotherapy, a proportion of patients (29.7%) received induction chemotherapy. The heterogeneity precludes accurate stratification and selection of patients of highest-risk of developing subsequent disease recurrence justifying routine imaging in this subgroup of patients. Finally, as this was a historical cohort, there was a lack of p16/ HPV data. Patients with p16/ HPV-positive oropharyngeal cancer have a more favorable prognosis than their counter parts. As the majority of the cohort had oropharyngeal cancer, our results may be biased by inadvertent inclusion of this subgroup of patients with favorable prognosis and low risk of subsequent failures. In addition, some newer immunotherapy agents and stereotactic radiotherapy were not available as a treatment option for the majority of our cohort who had a relapse. Nevertheless, this study, to our knowledge, is the largest cohort study to report on the utility and cost of surveillance imaging in patients with head and neck cancer treated definitively with radiotherapy in the contemporary era. Our cohort had long-term follow-up which provided us with ‘real-life’ estimates of recurrences and its presentation, salvage treatments, and successful salvage treatments. Furthermore, as not all patients are compliant with follow-up strategy suggested by their treating team, we collected data on every imaging that each patient had to provide a realistic estimate of cost of imaging for detection of a salvageable asymptomatic recurrence.

Conclusions

Our study showed that the rate of disease recurrence after successful treatment with definitive radiotherapy is low, with the majority of recurrences occurring in the first 2 years after treatment. Although recurrence was detected earlier with imaging, there was no significant difference in survival between those who had asymptomatic recurrence detected via imaging compared to those who had a recurrence detected clinically. Surveillance imaging in asymptomatic patients without clinically suspicious findings beyond 2 years requires judicious consideration and clinical acumen by the responsible clinician. Careful and thorough clinical assessment and examination including a nasoendoscopy assessment rather than imaging is recommended for long-term follow-up, reserving imaging to be performed at the clinician’s discretion and judgment.