Restaging FDG-PET/CT scan in recurrent cutaneous SCC: Diagnostic performance and prognostic significance

Abstract

Background:

There are no specific recommendations for FDG-PET/CT in assessing recurrent cutaneous squamous cell carcinoma (cSCC).

Objective:

To evaluate FDG-PET/CT in recurrent cSCC.

Methods:

FDG-PET/CT scans were retrospectively reviewed; sites of abnormal uptake were noted and correlated with biopsy/histopathology where available; follow-up imaging or clinical data in others. Comparison with available CT/MR was performed. Prognostic significance of PET/CT parameters was evaluated. PET/CT-based change in management was recorded.

Results:

115 FDG-PET/CT scans were analyzed in 100 consecutive cSCC patients. Of these, 96 (84%) scans were positive for recurrence and 25 showed distant metastases. PET/CT detected unsuspected disease sites in 39/115 scans (34%): locoregional disease, 14; distant metastases, 11; both, 8; additional local cutaneous disease, 5; and second malignancy, 1. Comparison of 78 PET/CT scans with available CT/MR demonstrated 37 additional abnormalities on 23 PET/CT scans, predominantly including skin/subcutaneous lesions and nodes. PET/CT led to change in management in 28% of patients. On univariate/multivariate analysis, increased number of FDG-positive lesions and lung metastases on PET/CT were associated with increased risk of death/disease progression.

Limitations:

Retrospective study.

Conclusions:

FDG-PET/CT was sensitive in detecting recurrent disease in cSCC, led to change in management in 28% of patients, and proved to be of prognostic value.

Capsule summary

1. Evidence guiding the use of imaging in evaluation of recurrent cutaneous squamous cell carcinoma is limited.

2. Our results demonstrate that (18F) 2-Fludeoxuglucose positron emission tomography/computed tomography has high sensitivity in the detection of recurrent disease, impacts patient management, and shows potential to be of prognostic value.

INTRODUCTION

The incidence of cutaneous squamous cell carcinoma (cSCC) has increased over the past few decades (1); approximately 3.3 million patients were diagnosed with keratinocyte carcinomas from 2006–2012 and 38.2% were treated for invasive cSCC (2,3). Management of cSCC varies based on several associated risk factors such as size, location, pathology of primary, timing of presentation, nodal involvement, and distant metastasis (4–6). Approximately 30–50% patients develop another SCC within five years after treatment, while <5% develop distant metastases (7–9).

Post-treatment follow-up is typically clinical and includes interval history and physical examination of skin and regional lymph nodes. Computed tomography (CT) with contrast may be indicated to evaluate regional nodes and distant sites in patients at risk for metastasis, while magnetic resonance (MR) imaging is used for deep soft tissue assessment or perineural spread (4,10–12). However, CT/MR may be inadequate for assessment of small tumors and nodes, and have limited field of view (4). The role of positron emission tomography/computed tomography (PET/CT) is currently unclear and may be considered based on clinician discretion to evaluate for distant metastasis.

(18F)-Fludeoxyglucose (FDG) PET/CT is reported to be of value for recurrence detection in non-cSCC skin malignancies (13,14). However, only few studies in literature with small patient cohorts have evaluated its role in cSCC (15–17). Limited data, in the form of few isolated case reports, is available to describe its clinical significance in detecting recurrent cSCC (18–22). In this study, we analyzed the diagnostic and prognostic performance of FDG-PET/CT in restaging cSCC patients.

METHODS

An IRB-approved retrospective study was performed in compliance with HIPAA regulations in patients with cSCC, treated at Memorial Sloan Kettering Cancer Center (MSKCC) from 2002–2016, who underwent restaging FDG-PET/CT scans. Patients with non-cutaneous primary, anogenital SCC, and cSCC in situ were excluded. The final cohort comprised 100 consecutive patients who had received prior treatments either at MSKCC or outside institutions. Electronic medical records were reviewed for clinico-pathologic data.

FDG-PET/CT Scanning

Patients injected with 370–555 MBq of FDG, following 6 hours of fasting and baseline blood glucose <200 mg/dl, underwent scan acquisition 60–90 minutes post-injection, from vertex-to-toes (73 scans), vertex-to-mid-thigh (40 scans), and eye-to-mid-thigh (2 scans). Low-dose CT with oral contrast was obtained. Attenuation-corrected images were reviewed on PACS workstations (AW suite2.0; GE Healthcare, Chicago, IL).

Image Analysis and Data Collection

PET/CT scans were reviewed by two independent NM physicians (SM, NPT). Focal areas of increased FDG uptake above the normal background, excluding physiologic sites, were considered suspicious and noted. Size and maximum standardized uptake value (SUV) of lesions were measured. FDG-avid suspicious foci were correlated with pathology, wherever available, or clinical/imaging follow-up (≥6 months). Imaging findings were classified as positive for recurrence/metastasis if confirmed by either positive histopathology from biopsies/resections or presence of detectable lesion at corresponding site on follow-up imaging showing increase in SUV or size. Comparison of PET/CT scans and CT/MR findings for limited regions/areas scanned was performed in 78 cases. All scans were acquired within one month of each other and in up to two-thirds of patients CT/MR was performed prior to PET/CT imaging with a mean interval of 12 days (range: 1–25 days) between the two.

Statistical Analysis

Sensitivity, PPV, and accuracy with 95% confidence intervals (CI) were calculated using pathology as the reference standard. Overall survival was defined from PET/CT scan to date of death; alive patients were censored at last follow-up. Time to progression (TTP) was defined from PET/CT scan to date of disease progression; patients with no progression recorded were censored at last follow-up. Kaplan-Meier curve was used to estimate survival rates. Log-rank test and Cox-proportional hazards model were used for univariate/multivariate analysis to assess prognostic value of PET parameters (p-value <0.05 considered statistically significant) including scan positivity, lesion count, SUV, lesion size, and site of metastasis. Adjusted hazard ratios (HR) are presented along with 95% CI. Statistical analysis was performed using R v3.5.0.

Management Change Assessment

The original treatment plan based on clinical examination and anatomic imaging was considered as the pre-PET plan. Both pre- and post-PET treatment plans were discussed with an independent clinician (CB) to understand influence on patient management, including inter-modality change (from surgery to radiation therapy (RT) or chemo-radiation; RT to chemotherapy; observation to surgery or chemotherapy), addition of different modality (from only surgery or only RT to combined surgery and RT) and intra-modality changes (increase in extent/dose of radiation or extent of surgical field).

RESULTS

A total of 115 restaging FDG-PET/CT scans in 100 consecutive cSCC patients were analyzed. Patient and scan characteristics are described in Table 1.

Table 1.

Patient and scan characteristics.

Number of patients	100
Gender (M:F)	84 (84%):16 (16%)
Mean age at the time of restaging FDG scan	74y (36–91)
Site of primary disease
Head and neck	87 (87.0%)
Trunk	6 (6.0%)
Extremities	7 (7.0%)
Number of FDG-PET/CT scans	115
Scan indications
Clinical suspicion (palpable node or visible skin tumor)	84 scans (n=77)
Biopsy-proven disease	11 scans (n=9)
Suspicious finding on CT/MR	9 scans (n=9)
Surveillance/post-operative site evaluation	11 scans (n=11)
Median duration (range) between initial diagnosis and restaging PET/CT, for each indication
Clinical suspicion	15.2 months (2.1–109.8 months)
Biopsy-proven disease	61.3 months (0.8–125.7 months)
Suspicious finding on CT/MR	60.8 months (0.6–186.3 months)
Surveillance/post-operative site evaluation	5.0 months (1.1–15.1 months)
Scans performed/patient
1 scan/patient	86
2 scans/patient	13
3 scans/patient	1
Treatment received after detection of recurrence
Surgery	23
Radiation therapy	18
Surgery + radiation	28
Chemotherapy ± surgery	15
Chemoradiation ± surgery	19
Unknown	12

Findings on FDG-PET/CT

FDG-positive foci were noted in 104 of 115 PET/CT scans while 11 scans were negative. Pathology-proven recurrent disease with or without distant metastasis was confirmed for at least one site in 96 scans (96/115; 84%) (Figure 1). Local recurrence only at/near the primary site or within post-operative cavity occurred in 19 scans (19/96; 20%), with regional node in 52 scans (52/96; 54%) while distant metastatic disease involving one or more sites was seen in 25 scans (25/96; 26%). Distant metastatic sites (pathology-proven in 15/25 scans) included bones, distant lymph nodes, lungs, ethmoid sinus, pericardium, muscular deposits, liver, bowel/peritoneum, and brain.

Figure 1.

A 75-year-old man with primary cSCC of scalp presented with swelling in right parotid gland, suspicious for recurrence and underwent FDG-PET/CT scan. Maximal intensity projection images demonstrated avidity in right parotid node measuring 1.2 × 1.0 cm (thick arrow, SUV 9) with unsuspected additional FDG-avid disease sites including multiple scalp lesions (thin black arrows), and right axillary lymph node measuring 1.5 × 0.8 cm (dashed arrow, SUV 10.8).

A total of 508 imaging abnormalities were recorded on either FDG-PET/CT or CT/MR, of which 482 foci were FDG-positive; predominantly including nodal (41%; 199/482), cutaneous (20%; 98/482), subcutaneous (8%; 39/482), osseous (10%; 47/482), pulmonary (6%; 28/482), pleural (7%; 34/482), and other sites (8%; 37/482). Nodes were located in the head and neck (77%; 154/199), thorax (13%; 25/199), and abdominopelvic region (10%; 20/199).

The overall mean SUV±SD (range) for FDG-positive pathology confirmed that recurrent disease was 8.7±5.8 (1.8–29.1) and for false-positive foci, 4.2±2.0 (2.2–9.1). Mean SUV specifically for FDG-positive recurrent nodes was 9.5±5.8 (1.8–5.0), for nodes ≤ 1 cm was 6.3±4.3 (1.8–22.8), and for nodes ≤ 0.5 cm was 2.7±1.8 (1.8–6.3). Mean SUV for distant metastases was 8.4 (range 2.3–20.1); specifically for lung metastases, it was 7.5±6.8 (2.3–20.1).

Overall mean size±SD (range) of FDG-positive nodes was 1.3±0.9 cm (0.4–8.3 cm) with 104/199 nodes (52%) ≤1.0 cm in short axis. Mean and median size of pathology-positive FDG-positive nodes was 1.6 cm and 1.3 cm (0.4–8.3 cm), respectively; and of distant nodes, 1.6 cm and 1.0 cm (0.8–4.5 cm) including axillary in four scans and mediastinal nodes (paratracheal, hilar, subcarinal) in three scans in patients with primary in head/neck.

Pathology Correlation

Of 508 abnormalities, histopathology was available for 212 and only follow-up clinical/imaging correlation for 256 imaging abnormalities, with no method of disease confirmation in 40. Considering pathology as the gold standard for cancer detection, calculated sensitivity, PPV, and accuracy of FDG-PET/CT scan was 99%, 94%, and 94%, respectively (Table 2). Twelve FDG-positive foci, negative for disease on pathology, included 7 nodes in head/neck region (Warthin’s tumor in 1, polymorphous lymphocytes in 1, reactive nodes in 4, and granulomatous in 1), 3 granulomatous mediastinal nodes (1 scan), and 2 cutaneous/subcutaneous abnormalities.

Table 2.

Diagnostic performance of imaging modalities by disease sites – using histopathology and imaging/clinical follow-up as standard of reference (along with 95% confidence intervals).

Lesion	No. of lesions	Sn (%)	Sp (%)	PPV (%)	NPV (%)	Accuracy (%)
LUNG
PET/CT	26	96 [80–100]	100 [2–100]	100 [86–100]	50 [1–99]	96 [80–100]
CT	17	100 [80–100]	NA	100 [80–100]	NA	100 [80–100]
BONE
PET/CT	45	100 [92–100]	NA	100 [92–100]	NA	100 [92–100]
CT/MR	14	64 [35–87]	NA	100 [66–100]	0 [0–52]	64 [35–87]
NODE
PET/CT	196	98 [95–100]	23 [10–42]	88 [82–92]	70 [35–93]	87 [81–91]
CT/MR	120	92 [85–97]	12 [2–38]	87 [80–93]	20 [3–56]	82 [74–88]
SKIN
PET/CT	86	100 [95–100]	0 [0–46]	93 [85–97]	NA	93 [85–97]
CT/MR	41	77 [61–89]	100 [16–100]	100 [88–100]	18 [2–52]	78 [62–89]
OVERALL (pathology as standard of reference)
PET/CT	212	99 [97–100]	14 [2–43]	94 [90–97]	67 [9–99]	94 [90–97]
*CT/MR	122	92 [86–96]	12 [0–53]	94 [88–97]	10 [0–45]	87 [80–92]

^*Specific regions acquired on CT/MR; Sn=Sensitivity, Sp=Specificity, PPV= Positive predictive value, NPV= Negative predictive value

Follow-up imaging confirmed recurrence in 218 FDG-positive foci. Nineteen FDG-positive sites, negative for disease on follow-up, included 7 cutaneous/subcutaneous sites and 12 nodes. Six abnormalities seen on CT/MR, FDG-negative sites, showed progression on follow-up imaging, including 3 nodes (2 cervical, 1 common iliac), 2 brain lesions, and 1 lung nodule.

Correlation with Anatomic Imaging

Correlation with CT/MR, available for 78 scans in 73 patients (73%; 73/100), showed concordant results in 48 scans. FDG-PET/CT detected 37 additional sites in 23 scans, negative on CT/MR, including 13 skin/subcutaneous (scalp, ear, nose, and post-operative site), 13 nodes (cervical, supraclavicular, parotid, retrocrural), 2 bowel, 4 muscle, and 5 osseous sites. Of 37, 30 lesions were positive for recurrence on either pathology or follow-up; 5 sites (4 nodes and 1 skin) with no correlation, and 2 skin foci negative for disease.

CT/MR identified 14 abnormalities that were FDG-negative in 10 scans, including 5 perineural/intracranial disease sites, 1 subcutaneous, and 8 nodes. Of 14, 7 sites (5 intracranial and 2 nodes) were positive for recurrence on pathology/follow-up imaging; 3 sites had no correlation, and 4 nodes were negative on follow-up. Overall accuracy of PET/CT compared to CT/MR was 87% vs. 82% in nodal metastases, 100% vs. 64% in osseous metastases, and 93% vs. 78% in cutaneous/subcutaneous tumors (Table 2).

Of note, additional foci were seen on 13 PET/CT scans that could not be assessed on available CT/MR due to limited/regional field of view including skin tumors (cheek, scalp, leg) and metastases in muscle, bones (pelvis, mandible), nodes (cervical, para-aortic, mediastinal, axillary), lung/pleura, and a second malignancy in epiglottis.

FDG-PET/CT-associated Change in Patient Management

Unsuspected findings (not known prior to FDG-PET/CT) were seen in 39/115 scans (34%), including additional cutaneous disease at/adjacent to site of treated primary in 5 scans, regional nodal disease in 14, distant metastases in 11, both locoregional and distant disease in 8, and unsuspected second malignancy in 1. PET/CT acquisition performed from vertex-to-toes or mid-thigh as opposed to conventional eye-to-thigh technique detected unknown tumors in scalp in 7 patients and lower extremities in 3 patients.

PET/CT resulted in inter-modality management change in 12 patients (12%; 12/100) and intra-modality changes or addition of modality in 16 patients (16%; 16/100) with overall change in management in 28/100 patients (28%) (Table 3). Additional surgical procedures included neck dissections (n=5), axillary dissection (n=1), VATS (n=2), and cutaneous scalp tumor excisions (n=2). Additional sites irradiated included sub-occipital nodes (n=1), humerus (n=1), epiglottis (n=1), philtrum (n=1), nasal-ala (n=1), lung (n=1), and neck (n=4).

Table 3.

Change in patient intended management plan post-FDG-PET/CT.

Pre-FDG-PET/CT		Post-FDG-PET/CT
Pre-PET/CT treatment plan	n	n	Post-PET/CT treatment plan	Change
Radiotherapy (RT) only	6	1	RT field ext	Radiation field extension
		1	CT only	RT stopped; CT added
		1	RT+CT	CT added
		1	RT field ext+Sx	Radiation field extension & excision for new primary lesions
		1	RT boost+CT	Radiation boost and CT added
		1	RT+Sx	Sx added
Surgery (Sx) only	14	3	Sx field ext+RT	RT added and additional neck dissection
		1	Sx field ext+RT	RT added and additional parotidectomy
		1	Sx field ext	Additional parotidectomy, MRND, additional skin lesion excision
		2	Sx field ext	Additional skin lesion excision
		3	RT+CT	Added RT & CT; Sx stopped
		1	RT	Added RT; Sx stopped
		3	CT	Added CT
Surgery and radiation	2	1	Sx+RT field ext	Radiation field extension for new lesion
		1	RT+CT	Added CT; Sx stopped
Observation	6	2	Sx	Added Sx
		1	Sx+CT	Added Sx+CT
		1	CT	Added CT
		1	RT+CT	Added RT and CT
		1	RT+CT	Added RT and CT

RT – Radiotherapy, field ext – field extension, Sx – Surgery, CT – Chemotherapy, n – number of patients.

Patient Outcome

Fifty-nine patients died by the time of analysis. Median follow-up for the remaining patients was 44months (range:0.5–99months). Median survival time was 37.4months (95%CI:20.5–55.3months). Re-occurrence of disease in the follow-up post-treatment period after PET/CT was seen in 71/100 patients (71%). The remaining 29/100 patients (29%) did not develop disease until the last follow-up. Median TTP was 12.3months (95%CI:9.0–16.7months).

Cox Regression Models and Overall Survival

Among the SUV threshold tested, none led to a difference in OS or TTP. Based on log-rank test results, not surprisingly, TTP was found to be longer for patients with negative PET/CT than positive ones (p=0.04; Figure 2). Multivariate analysis demonstrated significantly increased risk of death with the number of lesions (HR=1.58 [1.00–2.49] for a 5-lesion increase, p=0.05) and lung metastasis (HR=4.34 [1.54–12.30], p=0.006). Risk of progression also increased when the number of lesions increased (HR=1.35 [1.00–1.83] for a 5-lesion increase, p=0.05), with lung metastasis (HR=2.69 [1.22–5.93], p=0.01) as well as with bone metastasis (HR=2.67 [1.23–5.83], p=0.01).

Figure 2.

Prognostic stratification of FDG-PET/CT positivity using Kaplan Meier curves. (A) time to progression; (B) overall survival.

DISCUSSION

The current role of FDG-PET/CT in follow-up of cSCC is not well-defined and, to our knowledge, this is the largest series reported to date, specific to evaluation of recurrent cSCC. In our study, we noted PET/CT was sensitive in detecting recurrence in 84% of the scans; three-fourths were locoregional, while 25% of cases showed distant metastases (most common site was bone, comprising 10% of all lesions with a median SUV of 7.9). Recurrent disease was generally highly avid, with 80% of lesions showing SUV>4. About 3.5% of lesions were less avid with SUV<2, but were confirmed recurrence based on pathology. PET/CT was sensitive in detecting small lesions, including 52% of nodes ≤1 cm in size that are non-pathologic on CT/MR and can otherwise be clinically impalpable. In 30% of patients with available CT/MR, PET/CT identified 37 additional abnormalities not apparent on anatomic imaging, predominantly small nodes and cutaneous/subcutaneous lesions.

False-positive FDG-PET/CT findings, confirmed on pathology, primarily included lymph nodes, mostly reactive with few showing granulomatous inflammation. For false-positive cutaneous/subcutaneous sites, in the absence of pathology correlation and clinical follow-up showing absence of disease, we hypothesize these were probably related to non-specific inflammatory/infectious etiology. False-negative PET/CT findings in 5 cases with perineural/intracranial disease can be explained by lower image resolution and presence of physiologic uptake in brain parenchyma that may mask the low-grade uptake of tumors.

Bone metastases, better evaluated on FDG-PET/CT than CT/MR, were either morphologically indistinct or were outside the regional field-of-view. Other small-sized metastases such as cutaneous/subcutaneous lesions, muscular deposits, and bowel/pericardial deposits were also more apparent on PET compared to CT/MR; PET/CT showed greater accuracy in demonstrating involvement of lymph nodes, bones, and cutaneous/subcutaneous sites.

Detection of unsuspected nodes, often clinically impalpable, can influence surgical management and may determine the radiation field. Unsuspected distant metastasis also warrants a change in management. In our study, FDG-PET/CT findings led to a change in management in 28% of patients. In 4 patients, where only RT was planned, a change in field or addition of chemotherapy was directed by PET/CT; in 11 patients, where only surgery was planned, a change in surgical dissection (n=7) or change to another modality (n=4) was affected by PET/CT.

Recurrent advanced cSCC confers a poor prognosis, also established by the short progression-free survival seen in our cohort. No previous studies have evaluated FDG-PET/CT scan in cSCC as a prognostic tool. Our data showed that presence of lung metastasis, bone metastasis, and a high number of FDG-positive foci are independent and significant prognostic factors that unfavorably influence progression-free survival. This implies the potential to use metabolic imaging for risk stratification of patients with recurrent cSCC. Identification of patients at risk of poor outcome may warrant the use of intensive therapeutic approaches. However, prospective studies will be needed to further establish the prognostic value of FDG-PET/CT imaging.

Few studies have demonstrated the use of FDG-PET/CT for detecting recurrent disease in cSCC that include small heterogeneous or mixed patient population with variable timing for imaging (16,23); staging FDG-PET/CT was found to be beneficial in changing management for 6–22% of patients (15,17,24). Our study is distinctive in analyzing FDG-PET/CT data in a reasonably large patient cohort, with the majority of scans performed for clinical suspicion of disease.

Limitations of this study include its retrospective nature and probable selection bias given the variable disease stage and patient presentation. Not all patients had correlative dedicated CT/MR imaging for comparative analysis with PET/CT. However, this aligns with the utilization of anatomic imaging in current clinical practice. Pathology correlation was only available for some lesions (42% had direct tissue correlation). In addition, our data does not reflect on cost-effectiveness of PET/CT in this setting.

While standard PET/CT recommendations for oncologic imaging suggest scanning from skull base to thigh, no special recommendations apply specifically to cSCC (25). In patients with head and neck cSCC, vertex-to-thigh imaging is critical for detecting recurrence involving the scalp or forehead as well as locoregional metastasis. Additional cutaneous tumors detected in the lower extremities with whole-body PET acquisition, while only in small numbers (3/73 whole-body scans), is an additional advantage. The routine acquisition of vertex-to-toe scans in these patients is therefore of value.

CONCLUSION

Our findings suggest that FDG-PET/CT allows for whole-body scanning, shows high accuracy in detecting cSCC recurrence, and may be the modality of choice for disease assessment. FDG-PET/CT adds incremental value to anatomic imaging by detecting unsuspected disease sites and is advantageous for physicians in selecting the optimum treatment strategy for patient management. Additionally, the number of lesions and site of metastasis detected on PET/CT can influence prognosis.

Abbreviations used:

FDG-PET/CT

[18F] 2-Fludeoxuglucose positron emission tomography/computed tomography

cSCC

cutaneous squamous cell carcinoma

CT

computed tomography

MR

magnetic resonance imaging

PFS

Progression-free survival

OS

Overall survival

GLUT-1

Glucose transporter 1

SUV

Standardized uptake value

NCCN

National Comprehensive Cancer Network

HIPAA

Health Insurance Portability Accountability Act

MIP

maximum intensity projection

PACS

picture archiving and communication system

AJCC

American Joint Committee on Cancer

TP

true positive

PPV

positive predictive value

TTP

Time to progression

NM

Nuclear Medicine

IRB

Institutional Review Board