Impact of neck PET/CT positivity on survival outcomes - visual and quantitative assessment: Results from ACRIN 6685

Abstract

INTRODUCTION:

FDG PET/CT was prospectively studied in 287 cN0 Head and Neck Cancer (HNC) patients in ACRIN 6685 and additional analysis of neck lymph node FDG uptake upon recurrence-free survival (RFS) and overall survival (OS) was performed.

METHODS:

Two hundred eight had analyzable data. Survival analysis was performed to compare RFS and OS based on neck FDG visual assessment (VA) and maximum standardized uptake value (SUV_max). For SUV_max, the optimal thresholds were calculated using conditional inference trees on a randomly selected 70% training dataset and validated using the remaining 30% of data. Kaplan-Meier curves with log-rank tests were generated for the patient groups based on VA and optimal SUV_max thresholds, and the hazard ratios (HRs) and 95% confidence intervals (CIs) were also calculated. Hypothesis testing was set at a significance level of 0.05.

RESULTS:

73.9% of bilateral cN0 and 50.0% of unilateral cN0 were alive at the end of the study with the remaining being dead or lost to follow up. OS median follow-up time was 24.0 months (IQR 15.8–25.3, range 0–37.0). 63.3% of bilateral cN0 and 42.5% of unilateral cN0 patients remained disease free during the study. RFS median follow-up time was 23.9 months (IQR 12.4–25.2, range 0–35.6). VA of necks by our panel of radiologists was significantly associated with RFS (HR [95% CI]=2.30 [1.10, 4.79]; P=0.02), but not with OS (HR [95% CI]=1.64 [0.86, 3.14], P=0.13). The optimal SUV_max thresholds were 2.5 for RFS and 5.0 for OS. For SUV_max assessment, applying the optimal thresholds to the 30% test data yielded HRs (95% CIs) of 2.09 ([0.61, 7.14]; P=0.23) for RFS and 3.42 ([1.03, 11.41]; P=0.03) for OS. The SUV_max threshold of 5.0 was significantly associated with RFS (HR [95% CI]=5.92 [1.79, 19.57]; P<0.001).

CONCLUSION:

Neck FDG uptake by VA is significant for RFS. An SUV_max threshold of 5.0 is significantly associated with OS and RFS.

Graphical Abstract

INTRODUCTION

18 fluoro-2-deoxy (FDG) positron emission tomography (PET) combined with computerized tomography (CT) has represented a significant advance in the diagnosis, staging, and surveillance of most malignancies (1). FDG PET/CT merges functional and anatomic imaging techniques to specifically localize areas of anatomic abnormality and glucose (FDG) hypermetabolism. Hypermetabolism of glucose is a characteristic of viable malignant cells which is leveraged to locate and characterize malignant tumors with precision (2).

ACRIN 6685 collected data from patients with T2-T4 squamous cell head and neck cancer (HNC) across all primary tumor subsites who presented for treatment with cN0 disease (3). Please refer to our original study for greater demographic detail of the study group (3). Study criteria included negative cross-sectional imaging of the neck within six weeks of treatment and a four-week or less pre-operative FDG-PET-CT scan on an ACRIN certified scanner at one of 24 approved sites (22 sites enrolled, Table S1). Patients, following successful enrollment, were subjected to surgery for their primary disease, including unilateral or bilateral neck dissection(s) as deemed indicated by local surgeons. Pre and post FDG-PET-CT operative plans were recorded and change of plan based on FDG-PET-CT scan results was noted (3). Pathology specimen data was also recorded (4,5).

Disease status within the neck of HNC patients is crucial information for staging and prediction of outcome (6). Failure to control HNC within the neck, such as under-treatment of a cN0, is a significant complication and linked to significant morbidity and mortality (7). FDG uptake on PET/CT within the lymph nodes of the neck may indicate advanced stage (III and IV) and portend a worse prognosis (8).

Comparison of FDG-PET-CT to recurrence-free survival (RFS) and overall survival (OS) was a planned secondary objective of the 6685 study. We analyzed survival data from ACRN 6685, which accrued subjects for almost six and a half years. Clinical follow-up reached a median of 2 years (3). We analyzed this data for RFS and OS. FDG uptake in lymph nodes was analyzed in two ways: visual assessment (VA) by the central reading of FDG enhancement and region of interest (ROI) analysis of lymph nodes, which were measured for a standardized uptake value (SUV) (9). Survival analysis was used to determine whether pretreatment neck uptake by VA or SUV had an impact on RFS or OS.

MATERIAL AND METHODS

Participants

A total of 287 participants were recruited from 24 sites (Data Supplement). All participants had a newly diagnosed, first-time HNC that was evaluated for surgery. Participants with fasting glucose level > 200 mg/dL were excluded. All participants provided written informed consent and sites received approval from their local investigational review boards. All data were anonymized to protect the identities of participants (3).

Eligible participants had HNCs of the oral cavity, oropharynx, or larynx. One or both sides of the neck planned for dissection were clinically N0. A N0 neck was defined as free of palpable lymph nodes and with a neck CT and/or MRI reporting neck lymph node sizes of less than 1 cm in greatest diameter, excepting 1.5 cm for submental-submandibular nodes (Ia and Ib), jugular digastric nodes (IIa), or spinal accessory nodes (IIb), and showing a lack of central lymph node necrosis in nodes of any size based on clinical best practice (4). Participants received a presurgical FDG-PET-CT scan to which the surgeon was initially blinded following the contrast-enhanced CT or MRI scan of the neck. Local surgeons devised surgical plans per local standard of care. Pre FDG-PET-CT surgical plan (blinded to FDG-PET-CT results) and post FDG-PET-CT surgical plans were formulated. Both plans were collected prospectively through questionnaires.

Imaging

FDG-PET-CT was performed with a dedicated head and neck (H&N) PET-CT using two-bed positions from the orbits to the upper thorax (top of the aortic arch) with arms down and with images acquisitions at 6 minutes per bed position. There were 247 eligible participants; 54 underwent H&N scans only, 30 underwent whole-body (WB) scans only, the remainder had both H&N and whole-body scans performed. Acquisitions were 2 to 5 minutes per bed position depending on scanner. PET-CT scanners were qualified by the ACRIN PET Core Laboratory. A post-filter with an FWHM (full-width at half maximum) in the range of 5 mm was recommended for the dedicated head and neck acquisitions. FDG was administered IV 60 minutes prior to imaging (3.7 – 7.4 x 10⁸ Bq). Participants’ glucose levels were measured immediately prior to FDG injection.

Image Processing

PET-CT images were sent to the ACRIN Core Laboratory for quality control and assessment. Images were read by a panel of board-certified nuclear medicine or nuclear radiology physicians (n=4) who had no knowledge of the clinical care and were blinded to diagnosis, local PET/CT scan results, and clinical history. The core readers were instructed to characterize the participanťs primary malignancy and lymph node metastasis and evaluate the PET/CT scans for distant metastasis. Diagnosis of the primary, lymph nodes, and distant disease was made using a five-point ordinal scale: definitely malignant (5), probably malignant (4), indeterminate (3), probably benign (2), and definitely benign (1).

Two core readers independently read each scan. A maximum SUV (SUV_max, dose, and weight calculation method) was required for the "hottest" lymph node for each nodal basin recorded as indeterminate, probably malignant, or definitely malignant. The SUV_max calculation was performed on commercial software (MIM software, version 5.2, Cleveland, OH). For VA, "positive" nodal uptake of FDG was defined as uptake greater than background and more than the blood pool. The core VA readings were used for the primary aim calculations of NPV. Any discordant readings by core readers were adjudicated by a third independent reader who picked one of the readings as the final VA.

Neck Dissection

Surgery was performed according to local standards of care in an "intent to treat" fashion. The study did not mandate neck dissections, which would over-ride local clinical decision-making. Therefore, participants enrolled in the study would have been offered an elective neck dissection independent of the existence of this study. However, the type of neck dissection, when performed, had to include levels defined as first and second echelons for a given primary tumor and a minimum of 3 defined nodal levels.

Either bilateral or unilateral planned neck dissection was acceptable, as long as one side of the neck planned for dissection was clinically N0 pre-operatively. Clear margins were defined at a 2 mm minimum. Surgeons had knowledge of PET-CT results to guide the inclusion of any additional suspected nodal disease seen on PET-CT prior to surgery. Surgeons prospectively reported any revisions of their surgical plans based on the PET-CT results compared to their initial operative plan.

Pathological Interpretation

Pathology was read at local accrual sites, directed by a protocol for tissue sectioning, and all nodal levels were submitted at the time of dissection as separately labeled specimens following international designation (4,5). Due to the importance of pathology to this trial, a randomly selected sample of pathology reports were subjected to central quality review with slide requisition and reading by a single head and neck pathologist. Pathology data was compared to FDG-PET-CT results (5).

Each neck level was submitted indicating its side and level of origin. Lymph nodes were grossly dissected by pathologists, serially sectioned, and embedded in paraffin blocks. At least one slice was taken through each lymph node with the number of slices depending on the size of the lymph node. Each slice was stained with hematoxylin and eosin and read by the local pathologist following the study protocol. The final pathology report contained a list of nodal levels submitted, the total number of nodes found in each level, and the number of nodes positive for metastatic squamous cell cancer (SCC). A positive neck result was defined as at least one node found positive for SCC in a side of the neck.

VA and SUV_max

FDG uptake observed in any lymph node greater than the blood pool was considered positive. Readers recorded the SUV_max of neck lymph nodes for each neck level.

Statistical Analysis

For participants who did not have a dedicated head and neck FDG PET/CT scan, a whole-body FDG PET/CT scan was used (defined as using the "best available" scan). Participants who had at least one neck with a positive VA were considered to have a positive overall VA; otherwise, the participant was considered as having a negative overall VA. The mean SUV_max of the two central readers was computed to create a "combined SUV_max" summary measurement for each nodal basin.

Participants' vital and disease statuses were collected annually for two years following surgery. All participants who had recurrent disease reported during the follow-up period were considered in the RFS analysis; participant death was not considered to be an event in the RFS analysis. Participants who did not have a status of recurrent disease, but had "neck" reported as a site of distant metastasis, were coded as having recurrence and were also considered events in the RFS analysis. Participants who did not have disease recurrence were censored by the date that they were last known to be alive.

Participants whose dates of death were known were considered as events in the overall survival (OS) analysis. If only the month and year of death were known, then the middle of the month was imputed for the day of death. Participants who did not die, or who died and had missing months and years of death, were censored by the date they were last known to be alive.

The association between the overall VA and each event (recurrence or death) was assessed using Pearson's chi-squared test. Participants' follow-up times were summarized by the median, interquartile range (IQR), and range. The Kaplan-Meier survival curve for the positive overall VA was compared to the negative using the log-rank test. For each outcome, the hazard ratio (HR) comparing the positive and negative overall VA (negative as reference) and its 95% Wald CI were estimated by fitting a univariate Cox proportional-hazards model. These analyses were repeated in subsets defined by the number of cN0 necks at enrollment (either one side or both sides). Eligible participants who received an FDG-PET/CT scan and had surgery were considered to be analyzable. For all survival analyses, the surgery date was used as the index date (day 0).

In addition to Kaplan-Meier curve and Cox modeling, we also used the approach of conditional inference tree to search for optimal threshold of SUV_max for both RFS and OS (10). To reduce overfitting, we randomly split our sample into 70% training and 30% test datasets, stratified by each outcome of interest, in which we used the training datasets to search for the optimal threshold and used the test datasets to evaluate the performance. Kaplan-Meier curves with log-rank tests were then generated for the patient groups based on the tree-defined optimal threshold, and the HRs and 95% CIs were also calculated. The optimal thresholds were then evaluated on the entire analysis set for subsets defined by the number of cN0 necks at enrollment.

Hypothesis testing was set at a significance level of 0.05; P-values were not adjusted for multiplicity. All analyses were performed using SAS 9.4 (SAS Institute, Cary, NC) and R version 4.1.2 (R Foundation for Statistical Computing, Vienna, Austria).

RESULTS

ACRIN 6685 participant characteristics were reported using a Standards for the Reporting of Diagnostic Accuracy Studies diagram (STARD) in Supplemental Figure 1 (3). A total of 287 patients were enrolled. Table 1 shows the breakdown of unilateral and bilateral cN0 necks at enrollment. VA study reported on a selected group from ACRIN 6685. RFS analysis (N=208): Median follow-up time = 23.9 months (IQR 12.4–25.2, range 0–35.6) and OS analysis (N=208): Median follow-up time = 24.0 months (IQR 15.8–25.3, range 0–37.0). Table 2 reports the results of VA and demonstrates significance for neck FDG uptake and recurrent disease (P=0.02).

Table 1 –

Eligibility summary

	Both sides cN0	One side cN0	Unknown	Total
	N (%)	N (%)	N (%)	N (%)
Eligible	207 (92.0)	40 (67.8)	0 (0.0)	247 (86.1)
Ineligible	18 (8.0)	19 (32.2)	3 (100.0)	40 (13.9)
Total	225 (100.0)	59 (100.0)	3 (100.0)	287 (100.0)

Table 2 –

FDG-PET/CT visual assessment result versus recurrence

Visual assessment	Recurrence			P-value
	Yes	No	Total
	N (%)	N (%)	N (%)
Positive	26 (22.8)	88 (77.2)	114 (100.0)	0.02
Negative	10 (10.6)	84 (89.4)	94 (100.0)

Pearson's chi-squared test.

Supplemental Table 1 lists the participating institutions. Supplemental Table 2 displays survival data. 73.9% of bilateral cN0 and 50.0% of unilateral cN0 were alive at the end of the study with the remaining being dead or lost to follow up. Supplemental Table 3 presents the disease status assessment summary. 63.3% of bilateral cN0 and 42.5% of unilateral cN0 patients remained disease free during the study. There was persistent disease, recurrent disease or unknown status for bilateral cN0 (3.9%, 14.5%, and 18.4%) and for unilateral cN0 (5.0%, 17.5%, 35.0%) respectively.

Supplemental Tables 4 and 5 display the recurrence data for the entire group, local recurrence, regional recurrence, and distant metastases. Regional recurrence was 43.3% for bilateral cN0 and 57.1% for unilateral cN0 and 45.9% for the study overall. Supplemental Table 6 presents data on new secondary HNC mucosal primaries discovered during the study. One new HNC was discovered (0.4%) and 36 (14.6%) had missing data.

(a). Visual Assessment Analysis

Figure 1 displays Kaplan-Meier curves for RFS by FDG-PET/CT VA method. P-value from the log-rank test: P=0.02. Figure 2 displays curves for RFS, stratified by VA result, for (A) both sides cN0 and (B) one side cN0. P-values from the log-rank test P=0.02 and P=0.54, respectively. Figure 3 demonstrates OS, stratified by VA results with a p-value from the log-rank test: P=0.13, hazard ratio (95% CI) = 1.64 (0.86, 3.14).

Figure 1 –

Kaplan-Meier curves for recurrence-free survival (RFS), stratified by FDG-PET/CT visual assessment result. Two years of follow-up. P-value from the log-rank test: P=0.02.

Figure 2 –

Kaplan-Meier curves for recurrence-free survival (RFS), stratified by FDG-PET/CT visual assessment result, for (A) both sides cN0 and (B) one side cN0. Two years of follow-up. P-values from the log-rank test: (A) P=0.02, (B) P=0.54.

Figure 3 –

Kaplan-Meier curves for overall survival (OS), stratified by FDG-PET/CT visual assessment result. Two years of follow-up. P-value from the log-rank test: P=0.13.

Tables 3 and 4 report the median follow-up time for the VA for RFS and OS, respectively. Positive VA for RFS and OS was 23.3 (10.7–25.1, 0–35.6) and 23.9 (13.6–25.3, 0–37.0) months respectively.

Table 3 –

Follow-up time for RFS analysis (see figure 1), stratified by FDG-PET/CT visual assessment result

VA	N	Median follow-up time (IQR, range), months
Positive	114	23.3 (10.7–25.1, 0–35.6)
Negative	94	24.1 (17.5–25.3, 0–34.6)
Total	208	23.9 (12.4–25.2, 0–35.6)

IQR, interquartile range; RFS, recurrence-free survival.

Table 4 –

Follow-up time for OS analysis (see figure 3), stratified by FDG-PET/CT visual assessment result

VA	N	Median follow-up time (IQR, range), months
Positive	114	23.9 (13.6–25.3, 0–37.0)
Negative	94	24.2 (22.8–25.3, 0–34.6)
Total	208	24.0 (15.8–25.3, 0–37.0)

IQR, interquartile range; OS, overall survival.

Supplemental Tables 7 and 8 display VA and recurrence for the bilateral cN0 and the unilateral cN0. Recurrence was P=0.02 and 0.57 respectively. Supplemental Table 9 presents study RFS of 24.0 (14.6–25.3, 0–35.6) and 12.3 (5.9–24.7, 0–27.4) months for bilateral cN0 and unilateral cN0 respectively. Supplemental Tables 10-12 present VA and survival overall, for bilateral cN0, and unilateral cN0. This data was not significant for prediction of death regardless of the clinical neck status at study enrollment. Supplemental Table 13 displays follow up time with VA for OS for bilateral and unilateral cN0 of 24.1 (20.9–25.4, 0–37.0) and 20.2 (9.2–24.8, 0–27.4) months respectively.

(b). SUV_max analysis

OS analysis:

The 70% training dataset contained 146 participants for the conditional inference tree analysis. The optimal threshold of 5.0 was identified. In the 30% test dataset (n = 62), there were 52 participants with SUV_max less than 5.0, and 10 participants with SUV_max greater than or equal to 5.0. The median follow-up time was 24.0 (IQR 13.6–24.9, range 0–30.1, Table 5). For this dichotomized SUV_max, the HR (95% CI) was 3.42 (1.03, 11.41); the association between SUV_max and OS was statistically significant (P=0.03, Figure 4A). There was a significant difference between OS in the subset of participants who were bilateral cN0 (P=0.002), but not for participants who were unilateral cN0 (P=0.09). (Supplemental Tables 14-17).

Table 5 –

Follow-up time for RFS analysis (see figure 4), stratified by FDG-PET/CT SUVmax result

SUVmax	N	Median follow-up time (IQR, range), months
Positive (≥ 1.8)	142	23.7 (10.7–25.1, 0–35.6)
Negative (< 1.8)	66	24.2 (17.5–25.5, 0–34.6)
Total	208	23.9 (12.4–25.2, 0–35.6)

IQR, interquartile range; RFS, recurrence-free survival.

Figure 4 –

Kaplan-Meier curves for the optimal SUV_max thresholds. RFS (A), the hazard ratio (95% CI) is 3.42 (1.03, 11.41); the log-rank test P-value is P=0.03. OS (B), the hazard ratio (95% CI) is 2.09 (0.61, 7.14); the log-rank test P-value is P=0.23.

RFS analysis:

The 70% training dataset contained 145 participants for the conditional inference tree analysis. The optimal threshold of 2.465 was identified. In the 30% test dataset (n = 63), there were 33 participants with SUV_max less than 2.465, and 30 participants with SUV_max greater than or equal to 2.5. The median follow-up time was 24.0 months (IQR 12.3–25.5, range 0–30.6, Table 6). For this dichotomized SUV_max, the HR (95% CI) was 2.09 (0.61, 7.14). Using this threshold, the association between SUV_max and RFS was not statistically significant (P=0.23, Figure 4B). However, after applying the 5.0 threshold that was identified in the OS analysis to the RFS endpoint, the HR (95% CI) was 5.92 (1.79, 19.57) and the association was significant (P<0.001). Using the 2.465 SUV_max threshold, there was a significant difference between RFS among participants who were bilateral cN0 (P=0.02), but not unilateral cN0 (P=0.38). (Supplemental Tables 18).

Table 6.

Follow-up time for OS (see figure 7), stratified by FDG-PET/CT SUVmax and number of cN0 necks

	Both sides cN0		One side cN0
SUVmax	N	Median follow-up time (IQR, range), months	N	Median follow-up time (IQR, range), months
Positive (≥ 1.8)	123	24.0 (14.6–25.3, 0–37.0)	19	20.9 (9.0–24.9, 0–26.1)
Negative (< 1.8)	53	24.4 (23.3–26.2, 1.7–34.6)	13	16.3 (9.5–24.3, 0–27.4)
Total	176	24.1 (20.9–25.4, 0–37.0)	32	20.2 (9.2–24.8, 0–27.4)

IQR, interquartile range; OS, overall survival.

DISCUSSION

Failure in the neck in the HNC patient is significant for patients and their surgeons (7–9). Failure in the neck can result from initial under-treatment of the neck (8). This is more likely to happen in the cN0 patient than the cN+ patient (7,8) Under treatment often arises from an under-appreciation for the incidence of occult nodal metastases, a lack of familiarity with HNC primary locations and their potential first and second echelon nodes, and/or a failure to detect sub clinical disease in the regional lymph nodes on evaluation (11,12). These are rationales for the pre-operative use of FDG-CT-PET and/or lymphoscintigraphy with sentinel node dissection (3,5,13–15). In addition, these data are used to more accurately stage the patienťs neck in advance of initiating treatment (15,16).

We reported that FDG-PET-CT had a high NPV for N0 HNC in a prospective, multi-center series (3,5,13). There were 2,424 nodal levels from 303 sides of necks with pre-PET-CT and post-PET-CT documentation of the surgery plan (5). SUV analysis was superior to VA regarding the calculation of NPV. There was a high degree of reader agreement for FDG-PET-CT of the N0 neck, reassuring that these findings could be widely applicable. This suggested that FDG-PET-CT could help the clinician decide on the best therapy for the clinically N0 neck in HNC (3,5,13). We went on to report a description of HNC nodal metastatic behavior as reflected by FDG-PET-CT with a pathology gold standard (5). A well-designed clinical trial is underway which will complement our results and test the outcome of omitting routine neck dissection using PET/CT and sentinel lymph node biopsy (HN006) (15,16).

Twenty one patients had a local recurrence of which 5/21 (23.8%) had a regional recurrence and 3/21 (14.3%) had a distant recurrence. When local control was achieved, recurrence was 32.4% regionally and 10% distant respectively. The bilateral cN0 advantage persists for remaining disease free during our study period. The odds of persistent or recurrent disease over the study period were greatest for those presenting as unilateral cN+.

Neck FDG uptake by VA is significant for improved RFS in patients presenting with bilateral cN0. However, this does not hold true for unilateral cN+ patients. This point is clinically significant as not all HNC patients present as bilateral cN0. Primaries known for bilateral nodal metastases or that encroach the midline may be cN+ unilaterally, and yet the contralateral N0 neck requires investigation and possible treatment.

An SUV_max threshold of 5.0 is significantly associated with OS and RFS. Since we did not a priori establish a threshold for outcome prediction in the protocol, we divided the dataset into derivation (70%) and validation (30%) sets. We used the derivation data sets to establish the optimum thresholds of SUV_max 5.0 and 2.465 for OS and RFS, respectively, and then tested the associations using the validation datasets. As elaborated in the result, the SUV_max 2.5 threshold for RFS was not significant.

Limitations:

The N0 has long been a diagnostic and therapeutic challenge for head and neck surgeons. Although this has been recognized, it has been managed by diverse strategies among various medical/surgical disciplines and around the world. New technologies such as FDG-PET and its data radiomics have been advanced as a promising means of managing the N0 (16–20).

The discovery of optimal thresholds for SUV_max was analyzed using only one dataset, although we split the sample into training and test datasets to reduce overfitting. In the future, external validation with independent datasets is suggested to further confirm our findings.

There are number of reports indicating clinical and prognostic values of metabolic tumor volume (MTV) and total lesion glycolysis (TLG) as more suitable prognostic parameters than SUV in many oncology. These other parameters to analyze the ACRIN 6685 data are being explored in other future projects (21, 22).

CONCLUSIONS

Neck PET/CT positivity by visual assessment of FDG is significant for prediction of RFS. An SUV_max of 5.0 in neck lymph nodes is a possible threshold for predicting OS. This, in combination with a high NPV for FDG-PET-CT, may be a valuable indicator for clinicians in determining whether additional therapy to the neck is required and a useful marker for predicting the possibility of regional failure and its resultant dismal prognosis.

Supplementary Material

Supplemental Figure 1

ACRIN 6685 study flow chart.

Table 11

Visual assessment and survival for bilateral N0.

Table 12

Visual assessment and survival for unilateral N0 (c).

Table 13

Follow up time with visual assessment for overall survival (OS) and neck disease (N+).

Table 10

Visual assessment and survival for overall cohort.

Table 9

Visual assessment and study recurrence free survival (RFS).

Table 8

Visual assessment and recurrence for the unilateral N0 (b).

Table 7

Visual assessment and recurrence for the bilateral N0.

Table 6

New secondary head and neck mucosal primaries discovered.

Table 5

Recurrence data for local recurrence, regional recurrence, and distant metastases.

Table 3

Disease status assessment summary.

Table 4

Recurrence data for the entire study cohort.

Table 2

Group follow up survival data.

Table 1

Participating institutions and lead radiologists and surgeons.

Table 18

Follow up time with SUV max for recurrence free survival (RFS) and neck disease (N+).

Table 17

SUV max and overall survival for unilateral N0.

Table 16

SUV max and overall survival for bilateral N0.

Table 15

SUV max and overall cohort survival.

Table 14

Overall survival and SUV max.

KEY POINTS.

QUESTION:

What can we examine in PET/CT data of the neck to predict recurrence free survival and overall survival in head and neck cancer patients?

PERTINENT FINDINGS:

Survival analysis upon ACRIN 6685 subjects compared RFS and OS based on neck FDG visual assessment and optimal SUV_max thresholds of 2.465 and 5.0 for RFS and OS, respectively. Visual assessment of necks was statistically significant for reduced RFS.

IMPLICATIONS FOR PATIENT CARE:

This finding will provide more insight into the prognostic value of the visual assessment and SUV_max thresholds in cN0 patients undergoing pre-therapeutic PET/CT.