Eur J Nucl Med Mol Imaging DOI 10.1007/s00259-013-2372-4

ORIGINAL ARTICLE

Detection of underlying malignancy in patients with paraneoplastic neurological syndromes: comparison of 18F-FDG PET/CT and contrast-enhanced CT N. Schramm & A. Rominger & C. Schmidt & J. N. Morelli & C. Schmid-Tannwald & F. G. Meinel & M. F. Reiser & C. Rist

Received: 21 September 2012 / Accepted: 12 February 2013 # Springer-Verlag Berlin Heidelberg 2013

Abstract Purpose To determine the value of combined 18F-FDG PET/CT with diagnostic contrast-enhanced CT (CECT) in detecting primary malignancies and metastases in patients with paraneoplastic neurological syndromes (PNS) and to compare this with CECT alone. Methods PET/CT scans from 66 patients with PNS were retrospectively evaluated. Two blinded readers initially reviewed the CECT portion of each PET/CT scan. In a second session 3 months later, the readers analysed the combined PET/CT scans. Findings on each study were assessed using a four-point-scale (1 normal/benign; 2 inconclusive, further diagnostic work-up may be necessary; 3 N. Schramm (*) : C. Schmid-Tannwald : F. G. Meinel : M. F. Reiser : C. Rist Institute for Clinical Radiology, Ludwig-Maximilians-University Hospital Munich, Marchioninistrasse 15, 81377 Munich, Germany e-mail: [email protected] A. Rominger Department of Nuclear Medicine, Ludwig-Maximilians-University Hospital Munich, Munich, Germany C. Schmidt Department of Neurology, Ludwig-Maximilians-University Hospital Munich, Munich, Germany J. N. Morelli Department of Radiology, Texas A&M Health Sciences Center, Temple, TX, USA

malignant; 4 inflammatory). Sensitivity and specificity for malignant findings were calculated for PET/CT and CECT. Interreader agreement was determined by calculating Cohen’s kappa. Pooled data from clinical follow-up (including histopathology and follow-up imaging, median followup 20.0 months) served as the reference gold standard. Results Both readers classified 12 findings in ten patients (15 %) as malignant on the PET/CT scans (two patients had two primary tumours). One such imaging finding (suspected thymic cancer) was false-positive (i.e. benign histology). The most common tumours were bronchial carcinoma (n= 3), lymph node metastases of gynaecological tumours (n=3) and tonsillar carcinoma (n=2). Three of 12 findings (25 %) were not detected by CECT alone (cervical carcinoma, lymph node metastasis and tonsillar carcinoma). In a perpatient analysis, sensitivity and specificity for malignant findings were 100 % and 90 % for PET/CT and 78 % and 88 % for CECT. In 24 % (reader 1) and 21 % (reader 2) of the patients, the PET/CT findings were inconclusive. Of these findings, 57 % (reader 1) and 56 % (reader 2) were only diagnosed with PET (e.g. focal FDG uptake of the thyroid, gastrointestinal tract and ovaries). On follow-up, none of these findings corresponded to malignancy. Overall agreement between the two readers was excellent with a Cohen’s kappa of 0.95±0.04 (p<0.001) for PET/CT and 0.97±0.03 (p<0.001) for CECT alone. Conclusion In this cohort of patients with PNS, PET/CT exhibited improved detection of underlying malignancy versus CECT alone. While hybrid imaging produces a greater number of inconclusive findings, sensitivity is increased for the detection of head and neck and gynaecological malignancies as well as metastatic lymph node involvement.

Eur J Nucl Med Mol Imaging

Keywords PET/CT . Paraneoplastic neurological syndrome . Tumour detection . Contrast-enhanced CT

Introduction Paraneoplastic neurological syndromes (PNS) are a rare group of neurological disorders that occur in less than 1 % of cancer patients. PNS were initially described as neurological syndromes of unknown cause, associated with cancer but not attributable to a definite aetiology such as metastases, infections or treatment side effects [10, 13, 31]. Since 1985 several onconeural antibodies have been characterized which are associated with PNS and are directed against neural antigens expressed by malignant neoplasms [11, 31]. Most PNS are now thought to relate to immune-related effects of cancer on the nervous system [10, 25]. In 2004 the EFNS (European Federation of Neurological Societies) task force published diagnostic criteria for PNS [13]. The authors defined “classical” PNS such as limbic encephalitis, subacute cerebellar degeneration and subacute sensory neuronopathy as well as “wellcharacterized” onconeural antibodies such as anti-Hu and anti-Yo. Based on descriptions of classical and nonclassical PNS, the presence and absence of onconeural antibodies and other criteria, PNS can be classified as “definite” or “possible” PNS. Typical cancers associated with PNS include, for example, small-cell lung cancer (SCLC) and ovarian or breast carcinoma [25, 26, 31]. In most of patients, neurological symptoms precede symptoms related to the primary cancer, sometimes by years [27]. Therefore, early detection of the underlying malignancy is critical as allows earlier and specific treatment of the tumour which is the best approach facilitating stabilization of the neurological syndromes and improving prognosis [16, 30]. Imaging plays an important role in detecting underlying malignancy in PNS. If well-characterized antibodies such as anti-Hu are detected, the type of antibody present can guide imaging-based screening for the tumour [27]. For example, anti-Hu-antibodies are most often associated with SCLC, and their detection would warrant dedicated imaging of the thorax. However, the majority of tumours associated with PNS are small, often with only lymphatic metastases. Furthermore, patients can also have PNS without detectable paraneoplastic antibodies [13]. In these patients morphological imaging, for example CT or MRI, may fail to detect the underlying tumour [9]. Several studies have demonstrated the added value of FDG PET in comparison to anatomical imaging [3, 18, 22, 23, 32] for this purpose. Sensitivities for detection of underlying malignancy in the setting of PNS have been reported to be as high as 90 %. However, these studies examined relatively small patient cohorts, and in some studies only patients with positive antibodies or negative conventional imaging

results were included. FDG PET is recommended by the EFNS task force for most tumours associated with PNS if morphological imaging is negative [27]. Hybrid imaging with PET/CT is playing an increasingly important role in oncological imaging [5]. To our knowledge, only two studies with 46 [4] and 24 [21] patients and a case report [12] have examined the role of combined PET/CT in patients with PNS. In these studies, the CT component of the PET/CT was an unenhanced examination. The aim of this study was to retrospectively determine the value of combined 18F-FDG PET/CT with a fully diagnostic contrast-enhanced CT (CECT) component in detecting primary tumours and metastases in patients with PNS and to compare this with CECT alone.

Patients and methods Patients This retrospective study followed the principles of the Declaration of Helsinki and subsequent amendments [1]. Approval of the local institutional review board was waived since all studies were performed in the course of standard clinical diagnostic work-up. Informed consent was obtained from all patients prior to the PET/CT examination. Included in the study were 66 patients (38 women, 28 men; mean age 60 years, range 21-87 years). The patients were referred for PET/CT because of suspected paraneoplastic neurological syndrome (PNS) from the Neurology Department of our University Hospital. PET/CT scans from January 2007 to July 2011 were retrospectively analysed. Of the 66 patients, 50 had one PET/CT examination for this indication, and 16 patients (24 %) had follow-up PET/CT studies, up to four subsequent examinations. The preliminary clinical diagnosis of PNS was arrived at by senior neurologists at our institution after comprehensive neurological examination and exclusion of other possible causes of symptoms. PNS was classified as either “possible PNS” or “definite PNS”. Of the 66 patients, 58 (88 %) were tested for the presence of paraneoplastic antibodies in the Neurology Department. The different PNS of the patients are summarized in Table 1. PET/CT PET/CT examinations were performed on a 64-detector row PET/CT scanner (Siemens Biograph 64; Siemens Healthcare, Erlangen, Germany) after injection of an average of 250 MBq 18 F-fluoro-2-deoxy-D-glucose (FDG). Patients were asked to fast for at least 6 h before the examination to ensure blood glucose levels below 150 mg/dl. Butyl scopolamine (20 mg) was administered intravenously to avoid first-pass uptake of

Eur J Nucl Med Mol Imaging Table 1 Clinical indications for PET/CT Suspected PNS

Number of patients

Limbic encephalitis Cerebellar syndrome (atrophy, cerebellitis) Myositis/myopathy Polyneuropathy Lambert-Eaton myasthenic syndrome Motor neuron disease Dementia Stiff-person syndrome

12 11 10 10 6 4 3 2

Cranial nerve palsy Others Total

2 6 66

FDG into smooth muscle. Additionally, 20 mg of furosemide was given intravenously to increase renal excretion of the tracer. PET/CT image acquisition was initiated approximately 60 min after injection of the radiopharmaceutical. After the diagnostic contrast-enhanced CT images had been acquired, a PET scan (five or six bed positions, field of view 11 cm, 144× 144 matrix, three-dimensional mode, 2.5 min per bed position) was performed. The CT data were used for attenuation correction. Of 66 examinations, 55 were performed in combination with a diagnostic CT scan (100–190 mAs, 120 kV, 2.5 mm collimation, pitch 1.5) from the skull base to the proximal femora after intravenous injection of iodinated contrast agent (Iomeprol 350 mg/ml; Bracco SpA, Milan, Italy; body weight adapted administration, 1.5 ml/kg body weight) at 2.5 ml/s. Initiation of this scan was delayed by 80 s in order to image during the portal venous phase. In 11 of the 66 patients in whom a diagnostic CT had already been performed prior to but within 2 weeks of the PET/CT scan, a low-dose unenhanced CT scan was obtained to minimize patient exposure to ionizing radiation. The noncorrected and attenuationcorrected PET images as well as PET maximum intensity projection images were available for review. CECT, PET and fused PET/CT images could all be analysed in the axial, coronal and sagittal planes. Image analysis A radiology resident (6 years’ experience; reader 1) and an attending PET/CT specialist (>12 years’ experience in anatomical and functional imaging; reader 2) independently reviewed the CECT scans, both blinded to the clinical data. In a second session 3 months later, the two readers independently analysed the PET/CT scans. After assessing the PET/CT scans in their entirety, the readers were asked to identify whether a given finding was present on the PET portion alone, on the CECT portion alone, or on both study

components. The readers assessed the CECT and PET/CT scans using the following four-point-scale: Grade 1 Unremarkable examination or imaging findings considered benign (e.g. liver cysts). Grade 2 Inconclusive PET or CT findings not considered malignant, but findings for which further diagnostic work-up would be recommended in the clinical routine. An example of this would be focally increased FDG uptake within the gastrointestinal tract. Grade 3 PET and CT findings highly suspicious for malignancy: PET criteria were areas of focally increased FDG uptake not related to physiological accumulation or an inflammatory process. CT criteria for malignancy were, for example, enhancing masses, signs of tumour necrosis, invasion of adjacent structures, lymph nodes with a short axis diameter >1 cm, or metastatic disease. Grade 4 Findings considered to be inflammatory by PET and CT, comprising entities such as pulmonary infiltrates. The anatomical localization of the morphological findings and their size in centimetres, as well as the distribution of the metabolic findings, were recorded. PET findings with increased FDG uptake were evaluated qualitatively (malignant, inconclusive, inflammatory). Differences between detection of findings on CECT and PET/CT were assessed. In the 11 patients with a lowdose PET/CT scan, the separate diagnostic CECT scan obtained less than 2 weeks prior to the PET/CT scan was reviewed in both sessions. Results were correlated with clinical data (including antibody status) from medical records up to 30 September 2011. In all patients with malignant PET/CT findings (ten patients with 12 findings), the final diagnosis of the primary tumour was made after histological examination. In two patients with metastatic disease from cervical and ovarian cancer, the presence of metastatic lymph nodes was confirmed by imaging follow-up. In patients with inconclusive findings, medical records were reviewed for the results of imaging follow-up or specific diagnostic work-ups such as colonoscopy. The median follow-up period in all patients was 20.0 months (range 3–57 months). Statistical analysis Statistical analysis was performed using SPSS for Mac version 20.0.0.1. Interreader agreement was determined for PET/CT and CECT separately by calculating Cohen’s kappa as well as the corresponding standard errors and levels of significance. P values <0.05 were considered statistically

Eur J Nucl Med Mol Imaging Table 2 Patient characteristics

Interreader agreement

Characteristic

n

Number of patients Total Men Women Age (years) Mean Range Positive paraneoplastic antibodies

Overall agreement between the two readers was excellent with a Cohen’s kappa of 0.95±0.04 (p<0.001) for PET/CT and 0.97±0.03 (p<0.001) for CECT alone. Agreement between the readers was perfect for findings classified as malignant and inflammatory. Two patients had PET/CT findings and one patient had CECT findings that were classified as inconclusive by reader 1 and as benign/ unremarkable by reader 2. Table 3 shows the patient-based assessment of the two readers.

66 28 38 60 21–87

Positive Negative Not tested Definite paraneoplastic syndrome

19 (28.8 %) 39 (59.1 %) 8 (12.1 %) 7 (10.6 %)

significant. For the calculation of sensitivity and specificity, we used all available clinical data including data from the initial PET/CT examination, follow-up imaging and histopathology as a pooled reference standard. Findings classified as malignant by CECT alone or combined PET/CT were compared against this reference standard to calculate sensitivity and specificity. This was performed both on a perfinding and on a per-patient basis.

Results A total of 66 patients with suspected PNS were referred from the Neurology Department for PET/CT (Table 1). The major groups of PNS were: limbic encephalitis (n=12), cerebellar syndrome (n=11), myositis/myopathy (n=10), polyneuropathy (n=10), and Lambert-Eaton syndrome (n= 6). Further details are provided in Table 2. Paraneoplastic antibodies were positive in 19 patients, negative in 39 patients, and not tested in 8 patients. PNS was neurologically categorized as “definite” in 7 patients. The median clinical follow-up period of the patients was 20.0 months (range 3– 57 months).

Table 3 Results of the patientbased assessment with PET/CT and CECT. Cohen’s Kappa PET/ CT: 0.95, Cohen’s Kappa CECT: 0.97) a

In one patient with two primary tumours only the lung cancer was detected with CECT, not the tonsillar cancer. Nevertheless, this patient was classified as “malignant” by CECT.

Malignant findings In ten of the 66 patients (15 %), 12 PET/CT findings were classified as malignant. There was perfect agreement between the readers (kappa=1). Two separate primary malignancies were found in two patients (one patient with both metastatic bronchial and bladder carcinoma, and another with peripheral bronchial and tonsillar carcinoma; see Fig. 1). The diagnoses made by imaging were bronchial carcinoma (n = 3; see Fig. 2), lymph node metastases of gynaecological cancers (breast, cervix, ovary; n=3), tonsillar carcinoma (n=2), cervical carcinoma (n=1), breast cancer (n=1), bladder cancer (n=1), and thymic cancer (n=1). Histology in the patient with suspicion of thymic cancer showed only stimulated T and B cells without evidence of malignancy. This was the only false-positive CECT or PET/CT finding (Fig. 3). The final clinical diagnoses in the remaining patients were in agreement with the PET/CT diagnoses and were confirmed histologically or on imaging follow-up. Three findings were detected only by PET/CT (the PET component) and not by CECT alone: cervical carcinoma, tonsillar carcinoma, and a lymph node metastasis of an ovarian malignancy (Fig. 1). Thus there was disagreement between the PET/CT and CECT results in 3/12 (25 %) cases. In the remaining 9/12 (75 %) cases, the relevant findings were detected by both CECT and PET. In one patient with tonsillar carcinoma cervical lymph nodes were only classified as malignant on the PET examination, and in another

PET/CT

CECT

Reader 1

Reader 2

Reader 1

Reader 2

Malignant

10 (15 %)

10 (15 %)

8 (12 %)*

8 (12 %)a

Inconclusive Inflammatory Unremarkable Patients total

16 (24 %) 4 (6 %) 36 (55 %) 66

14 (21 %) 4 (6 %) 38 (58 %) 66

9 (14 %) 4 (6 %) 45 (68 %) 66

8 (12 %) 4 (6 %) 46 (70 %) 66

Eur J Nucl Med Mol Imaging

a

b a

c

Fig. 3 False-positive PET/CT finding in a 24-year old male patient suffering from limbic encephalitis. a The CT image shows a nodular soft-tissue mass in the anterior superior mediastinum (white arrow). b The corresponding PET image shows patchy increased FDG uptake within the anterior superior mediastinum. Both the CT and PET findings were rated as malignant by the blinded reviewers. However, histology was benign (stimulated T cells)

d

Fig. 1 A 79-year-old male patient with limbic encephalitis and tonsillar cancer. a, b Additional value of PET imaging. a On the CT image no detectable soft-tissue asymmetry is present in the right tonsillar region to suggest the presence of malignancy (white arrow). b The PET image shows an area of focal hypermetabolic activity (black long arrow) within the right tonsil which corresponded to biopsy-proven tonsillar cancer. On the non-attenuation corrected images there are no metal artifacts. c, d A lung adenocarcinoma was also found. c A pulmonary nodule is present in segment 9 of the left lower lobe (short arrow). d This nodule demonstrates focal hypermetabolic FDG uptake (short arrow)

patient with bronchial and bladder cancer some nodal, bone, and soft-tissue metastases were likewise only detected by PET/CT. Further details are provided in Table 4. Positive paraneoplastic antibodies were identified in a patient with bronchial cancer, the patient with bronchial and tonsillar cancer, and in the patient with lymph node metastases from ovarian carcinoma (three of ten patients, 30 %). These patients and a patient with lymph node metastases from breast cancer were neurologically categorized as having a definite PNS (four of ten patients, 40 %).

a

b

b

Fig. 2 SCLC in a 58-year-old female patient with Lambert-Eaton myasthenic syndrome and positive anti-Hu antibodies. a The CT image shows an enhancing mass in the right hilum (arrow). b On the fused PET/CT images, the mass shows focal hypermetabolic uptake. c The

In the per-finding analysis, sensitivity and specificity for malignant findings were, respectively, 100 % and 91 % for PET/CT and 73 % and 89 % for CECT. In the per-patient analysis, sensitivity and specificity for malignant findings were, respectively, 100 % and 90 % for PET/CT and 78 % and 88 % for CECT. Inconclusive findings Reader 1 reported 21 inconclusive findings in 16 of the 66 patients (24 %). In five patients, there were two inconclusive findings. Of these 21 findings, 12 (57 %) were only diagnosed by PET and 3 (14 %) only by CECT. The most common findings were: focally increased FDG uptake in the gastrointestinal tract (Fig. 4), focally increased ovarian FDG uptake, focally increased thyroid FDG uptake or nodules and lymph nodes with increased FDG uptake. In contrast, reader 2 reported 18 inconclusive findings in 14 of the 66 patients (21 %). Five patients had two inconclusive findings. The aetiologies of the most common inconclusive

c coronal PET image shows additional hypermetabolic hilar and mediastinal lymph node metastases. In this patient, the mediastinal and hilar lymph nodes were also pathological according to CT criteria

M

F

F

M

3 (two 74 primary tumours)

67

78

24

79

50

4

5

6

7

8

9 (two 79 primary tumours)

F

M

F

Yes

No

No

No

No

Encephalitis

Yes

Polyneuropathy Yes

Cerebellitis

Cerebellar syndrome Limbic encephalitis

Polymyositis

Polymyositis

Dementia

No

F

64

2

Paraaortic foci

Right hilum, mediastinum, supraclavicular, right upper lobe

PET: location of focally increased FDG uptake

No

No

Only PET finding?

No

Yes

Tonsils symmetrical, CT not suspicious

Mass left lower lung lobe

Multiple enlarged lymph nodes left axilla

Focus left lower lung lobe. Focus right tonsil

Foci left axilla

Lymph node ventral to portal vein Lymph node ventral to (borderline; short diameter <10 mm) portal vein

Uterus and cervix not suspicious on Cervix CT Slightly nodular soft-tissue mass in the Upper anterior upper anterior mediastinum mediastinum/thymus

Yes

No

No

Yes

No

Yes

Yes

Not tested

Yes

No

No

True positive False positive

True positive

True positive

True positive

True positive

Assessment of PET/CT finding

Limbic encephalitis, (histology: stimulated T cells, benign) Lymph node metastases from True positive ovary cancer (histology, imaging follow-up) Lymph node metastases from True occult breast cancer positive (histology) Adenocarcinoma True of the lung positive Tonsil carcinoma (histology)

Breast cancer with lymph node metastases, lymphangitic carcinomatosis, and lung metastases (histology, imaging follow-up) Cervical cancer (histology)

Bronchial cancer with lymph node, bone, softtissue and lung metastases (histology, imaging follow-up)

Lymph node metastases from cervical carcinoma (histology, imaging follow-up) Bladder cancer with lymph node metastases

SCLC (histology)

Paraneoplastic Final clinical diagnosis antibodies? (method)

Dorsal bladder. No (some No Iliac chain, lymph retroperitoneal, nodes, mediastinal, bone, softsupraclavicular tissue foci metastases Mass central right lung, only Focus right central multiple lung nodules. detectable lung. Multiple Enlarged mediastinal and by PET) foci in both lungs. supraclavicular lymph Foci in the ribs nodes. Thoracic soft-tissue masses and thoracic soft tissue Multiple enlarged cervical, thoracic Multiple foci cervical, No No and abdominal lymph nodes. thoracic and abdominal. Multiple lung nodules. Suspicion of Foci in both lungs. lymphangitic carcinomatosis Patchy muscle uptake

Mass dorsal bladder. Enlarged lymph nodes iliac chain, retroperitoneal

Mass right hilum. Enlarged lymph nodes right hilum and mediastinum, supraclavicular. Nodule right upper lobe Two enlarged paraaortic lymph nodes

Yes

Lambert-Eaton myasthenic syndrome

58

1

F

Definite CECT finding PNS?

Patient no. Age Sex PNS (years)

Table 4 Patients with malignant PET/CT findings

Eur J Nucl Med Mol Imaging

True positive Tonsil carcinoma with lymph node metastases (histology) Cervical lymph node right

No (some Not tested lymph node metastases only on PET) Right tonsil

Asymmetry and mass with contrast enhancement right tonsil Enlarged right cervical lymph node No

Inflammatory findings Of the 66 patients, 4 (6 %) had findings classified as inflammatory by both readers. These findings were detected by both CECT and PET/CT. Two of these patients had pneumonia. One patient with a final diagnosis of CNS vasculitis had nonspecific enlargement of the mediastinal lymph nodes, and one patient with a clinical diagnosis of sarcoidosis had pulmonary nodules and enlarged hilar lymph nodes. The latter patient had positive paraneoplastic antibodies, and the PNS was categorized as “definite”. However, no evidence of malignancy was present on follow-up imaging. Unremarkable examinations According to reader 1, 36 patients (55 %) had an unremarkable PET/CT scan without any suspicious findings, and 45 patients (68 %) had a normal CECT scan. Reader 2 rated the PET/CT scan normal in 38 patients (38 %) and the CECT scan normal in 46 patients (70 %). The differences between PET/CT and CECT were due to the higher number of inconclusive findings diagnosed on the PET component (see above). Ten of the patients with unremarkable examinations (28 %) had positive paraneoplastic antibodies. No malignancies were detected subsequently in any of these patients on clinical follow-up. One patient with stiff-person-syndrome had a history of breast cancer and was clinically classified as having definite PNS. However, no malignancy was identified on subsequent follow-up.

M

Cerebellar syndrome

findings were similar to those identified by reader 1. Of these 18 findings, 10 (56 %) were only detected by PET and 2 by CECT alone (11 %). Further details are provided in Table 5. In no case with either reader did the CECT results clarify inconclusive findings seen by PET/CT. Although in some cases, CECT did alter the classification of a finding. For example, a nonspecific pulmonary FDG uptake was classified as “inflammatory” after a pneumonic infiltrate was detected on CT. Five of the patients with inconclusive imaging findings had positive paraneoplastic antibodies. One patient with positive Anti-Hu antibodies and definite PNS had a history of SCLC. However, in this patient and the other patients with inconclusive imaging findings, no primary tumour or metastatic disease was found during the course of clinical follow-up.

54

Discussion

10

Patient no. Age Sex PNS (years)

Table 4 (continued)

Definite CECT finding PNS?

PET: location of focally increased FDG uptake

Only PET finding?

Paraneoplastic Final clinical diagnosis antibodies? (method)

Assessment of PET/CT finding

Eur J Nucl Med Mol Imaging

PET and particularly PET/CT play an increasingly important role in oncological imaging and have demonstrated added diagnostic value in comparison to CT for the assessment of malignancies such as bronchial carcinoma, colorectal

Eur J Nucl Med Mol Imaging Fig. 4 Inconclusive finding in a 46-year-old female patient with dermatomyositis. The CT image (a) shows no abnormality, but the PET image (b) shows focal FDG uptake at the hepatic flexure of the colon. Subsequent colonoscopy was unremarkable

a

carcinoma, and lymphoma. Indications for PET and PET/CT include tumour detection, lesion characterization, staging, detection of tumour relapse, and therapy monitoring [5–7]. Detection of underlying malignancy in patients with PNS remains a diagnostic challenge. After the clinical onset of PNS, the diagnosis of occult malignancy is often delayed, sometimes for years [14, 24]. According to the EFNS task force criteria, PNS is definitively diagnosed when cancer develops within 5 years of the diagnosis of a classical PNS [13]. Many of the primary tumours are small, and occasionally only lymph node metastases of occult primary tumours are identified [9]. However, early detection of underlying malignancy is essential, as initiating treatment tailored to the primary tumour is the most effective means of stabilizing the PNS [30]. Depending upon the suspected causative neoplasm of the PNS, morphological imaging with CT has demonstrated

b

moderate sensitivity for tumour detection. For instance, thoracic CT exhibited a sensitivity of 83 % for detection of malignancy in patients with Lambert-Eaton myasthenic syndrome in a study by Titulaer et al. [28]. Likewise, sensitivity was 85 % for CT in detecting ovarian carcinoma in a study by Liu et al. comparing ultrasound, CT and MRI [19]. Due to the moderate and in other works even low sensitivities of purely anatomical imaging, several studies subsequently examined the efficacy of FDG PET in detecting malignancy in the setting of PNS [3, 15, 18, 22, 32]. The methodology and inclusion criteria of these studies were variable. The initial studies were small, including only antibody-positive patients, and showed very high sensitivities for PET. For example, Linke et al. [18] incorporated PET into the initial imaging work-up and compared CT and PET in 13 patients with positive onconeural antibodies. The authors found a sensitivity of 90 % for PET in the detection of underlying

Table 5 Patients with inconclusive findings

Inconclusive finding (n)

Total Patients with inconclusive findings PET and CT finding Only PET finding Only CT finding Malignancy after further diagnostic work-upa a

Median overall follow-up time 20.0 months.

Reader 1

Reader 2

Gut/stomach focal FDG uptake (4) Ovary FDG uptake/contrast enhancement (3) Focal thyroid FDG uptake/nodule (3) Lymph node with FDG uptake (3) Prostate focal FDG uptake (1) Tonsil asymmetrical FDG uptake (1) Lung infiltrate with focal FDG uptake (1) Mesentery focal FDG uptake (1) Skin thickening and FDG uptake (1) Hypervascular mass, vastus muscle (1) Hypervascular mass, breast (1) Questionable mass, pancreatic head/differential diagnosis partial volume effect (1) 21 findings in 16 patients 16/66 (24 %) 6/21 (29 %) 12/21 (57 %) 3/21 (14 %)

Lymph node with FDG uptake (4) Gut/stomach focal FDG uptake (3) Ovary FDG uptake/contrast enhancement (3) Focal thyroid FDG uptake/nodule (2) Small lung nodule with FDG uptake (1) Tonsil asymmetrical FDG uptake (1) Lung infiltrate with focal FDG uptake (1) Skin thickening and FDG uptake (1) Hypervascular mass, vastus muscle (1) Hypervascular mass, breast (1)

0

0

18 findings in 14 patients 14/66 (21 %) 6/18 (33 %) 10/18 (56 %) 2/18 (11 %)

Eur J Nucl Med Mol Imaging

malignancy versus only 30 % for CT, and the sensitivity for both modalities was 100 %. The majority of the early studies investigated the role of PET after conventional imaging was negative [3]. Other works included patients with negative antibody status but clinically suspected PNS. In these studies, PET findings suggestive of malignancy were found in 37 % (n=43) [23] and 23 % (n=80) [15] of patients, respectively. In a study by Patel et al. [22], PET had a sensitivity of 80 % and a specificity of 67 % (n=104) for the detection of underlying malignancy in patients with PNS. To the best of our knowledge, the present study is the largest series (n= 66) to date examining the efficacy of hybrid PET/CT incorporating CECT for the detection of underlying malignancy in patients with suspected PNS. Bannas et al. examined 46 patients [4] in their work, and a recent study by Matsuhisa et al. included 27 patients [21]. However, in both of these studies PET/CT imaging was performed with low-dose, nondiagnostic quality CT. In our study, in the per-patient analysis, sensitivity and specificity for malignant findings were 100 % and 90 % for PET/CT and 78 % and 88 % for CECT, respectively. The calculated values for sensitivity and specificity of PET/CT and CECT were based on the relatively small number of findings classified as malignant in our study and therefore have to be regarded as preliminary approximations. However, this is an inevitable consequence of the rarity of PNS. Since our study is the largest PET/CT study in this entity to date, the calculated test results are nonetheless important data. In the present patient cohort, 12 PET/CT findings in ten patients (15 %) were determined to be malignant by both readers. Interestingly, two patients were ultimately diagnosed with two synchronous primary tumours. One patient (patient 9) was diagnosed with lung adenocarcinoma and a synchronous tonsillar carcinoma, and another (patient 3) with synchronous bronchial and urinary bladder cancer. These findings highlight the potential of combined PET/ CT as a whole-body modality for detecting tumours in disparate body regions. To our knowledge, synchronous detection of two primary tumours in patients with PNS has not been described previously in the literature. One finding (1 of 12) false-positive for malignancy was identified by both readers on both the PET and CECT examinations. In this relatively young patient with symptoms of limbic encephalitis, imaging demonstrated a nodular soft-tissue mass with focally increased FDG uptake in the anterior superior mediastinum. Of note, previous studies have shown FDG PET to be helpful for the characterization of thymic pathology [17, 20]. However, in our patient histology revealed thymic hyperplasia. The most common true-positive malignant tumours were bronchial carcinomas (n =3), lymph node metastases of gynaecological tumours (n= 3) and tonsillar carcinomas (n=2). Three of the 12 malignancies (25 %) were only

detected by PET/CT (cervical carcinoma, lymph node metastasis of ovarian cancer and tonsillar cancer). We thus conclude that the PET component of the PET/CT has particular diagnostic value, increasing accuracy in the detection of head and neck and gynaecological malignancies as well as in the detection of metastatic lymph node involvement. These observations are in accordance with those of previous studies. In a study by Younes-Mhenni et al. [32], 12 of 14 patients with antibody-positive PNS and histologically proven tumours had PET-positive lymph node involvement, commonly within the mediastinum. Several review articles have confirmed the utility of PET/CT in the detection of head and neck and gynaecological cancers [8, 29]. As mentioned already, the present study evaluated PET/CT performed with diagnostic quality CECT, whereas previous studies in patients with PNS have only utilized hybrid imaging with unenhanced CT [4, 21]. In the present work, no malignant findings were detected by CECT alone. However, it is our belief that PET/CT imaging with a fully diagnostic CECT component has additional diagnostic value in PNS patients and is complementary to metabolic imaging. As has been shown in previous studies, we think that contrast-enhanced CT imaging improves anatomical lesion localization in regions of complex anatomy such as the head and neck region and within the abdomen and pelvis. Furthermore, lesion characterization can be improved by CT densitometry or by assessment of enhancement patterns following administration of contrast agent [2]. Bannas et al. concluded that the role of PET in the detection of underlying malignancy in PNS may have been previously overestimated [4]. For example, Linke et al. reported a sensitivity of 90 % for FDG PET and only 30 % for CT in their study in 13 antibody-positive PNS patients [18]. In contrast, Bannas et al. found a primary tumour in only 4 of 46 patients with PNS (8.7 %). In a more recent study, an underlying malignancy was detected by PET/CT in 5 of 27 of patients with PNS (18.5 %) [21]. In our study, the underlying tumour or tumours were identified in 9 of 66 patients with PNS (13.6 %) with two patients having two synchronous primary tumours. Even though the detection rate of the underlying malignancy in patients referred with suspected PNS is relatively low, in our opinion an initial examination with PET/CT (with diagnostic CT component) is indicated, as these patients are severely ill and detection of a tumour has crucial therapeutic and prognostic implications. However, further studies examining the relative cost-effectiveness of these approaches may be useful. According to previously published diagnostic criteria [13], PNS can be diagnosed without the presence of paraneoplastic antibodies. The 2011 EFNS Task Force guideline [27] recommends whole-body FDG PET only after conventional imaging was negative. In our study only three of ten patients (30 %)

Eur J Nucl Med Mol Imaging

with malignant tumour and PNS had positive paraneoplastic antibodies (two patients were not tested for antibodies). In a recent PET/CT study in 27 patients with PNS, underlying malignancy was detected in five patients, none of whom tested positive for paraneoplastic antibodies [21]. Similar to a 2007 study utilizing PET to assess PNS [15], our results support PET/CT as the initial imaging study in patients with clinically suspected PNS, regardless of their paraneoplastic antibody status. In distinction to previous studies, we did not classify PET/CT findings as strictly “positive” or “negative” but also included the category of “inconclusive findings”, defined as findings requiring further diagnostic work-up. For reader 1, 24 % of the patients had inconclusive findings and for reader 2, 21 % of the patients. Of 21 inconclusive findings, 57 % (reader 1) and 56 % (reader 2) were only detected with the PET component of the examination. The most common inconclusive findings were areas of focally increased FDG uptake in the gastrointestinal tract, thyroid, and ovaries. On follow-up none of these findings corresponded to malignancy. In our opinion, classifying PET/CT findings strictly as “positive/malignant” and “negative/benign” as has been done in previous studies is an oversimplification and fails to account for the realities of clinical practice. For example, in daily clinical routine an area of focally increased FDG uptake in the colon often leads to a subsequent colonoscopy. Our study had several limitations. The study was performed retrospectively. The number of patients with positive paraneoplastic antibodies was quite low (29 %). Even though this is the largest study utilizing primary PET/CT with a CECT component, an underlying malignancy was detected in only 13.6 % of the patients. Thus, only descriptive statistics could be applied as the number of patients with malignant findings was too small to assess statistical significance. The median clinical follow-up period in this study was 20.0 months. As several years often pass between the time of PNS onset and the detection of an underlying malignancy, data from longer-term longitudinal PET/CT follow-up of the patients in our study may prove interesting. Unfortunately, in this patient cohort there were not enough follow-up PET/CT scans performed to allow meaningful analysis. Conclusion In summary, PET/CT correctly detected underlying malignancies in 9 of 66 patients with PNS (13.6 %; one falsepositive finding), and exhibited improved lesion detection versus CECT alone. While hybrid imaging produces a greater number of inconclusive findings, sensitivity for tumour detection is increased especially with respect to head and neck and gynaecological malignancies, as well as metastatic lymph node involvement. The CECT component of the

PET/CT may provide exact anatomical localization as well as lesion characterization in contrast to unenhanced lowdose CT. In patients with PNS, detection of underlying malignancy is essential for treatment planning. Our findings suggest that hybrid imaging utilizing FDG PET and CECT should be the initial study of choice for detection of underlying malignancy in this patient group. Conflicts of interest None.

References 1. World Medical Association Declaration of Helsinki: ethical principles for medical research involving human subjects. JAMA. 2000;284:3043–5. 2. Antoch G, Freudenberg LS, Beyer T, Bockisch A, Debatin JF. To enhance or not to enhance? 18F-FDG and CT contrast agents in dual-modality 18F-FDG PET/CT. J Nucl Med. 2004;45 Suppl 1:56S–65S. 3. Antoine JC, Cinotti L, Tilikete C, Bouhour F, Camdessanché JP, Confavreux C, et al. [18F]fluorodeoxyglucose positron emission tomography in the diagnosis of cancer in patients with paraneoplastic neurological syndrome and anti-Hu antibodies. Ann Neurol. 2000;48:105–8. 4. Bannas P, Weber C, Derlin T, Lambert J, Leypoldt F, Adam G, et al. 18F-FDG-PET/CT in the diagnosis of paraneoplastic neurological syndromes: a retrospective analysis. Eur Radiol. 2010;20:923–30. 5. Bar-Shalom R, Yefremov N, Guralnik L, Gaitini D, Frenkel A, Kuten A, et al. Clinical performance of PET/CT in evaluation of cancer: additional value for diagnostic imaging and patient management. J Nucl Med. 2003;44:1200–9. 6. Basu S, Alavi A. Unparalleled contribution of 18F-FDG PET to medicine over 3 decades. J Nucl Med. 2008;49:17N–21N 37N. 7. Bomanji JB, Costa DC, Ell PJ. Clinical role of positron emission tomography in oncology. Lancet Oncol. 2001;2:157–164. 8. Cashman EC, Macmahon PJ, Shelly MJ, Kavanagh EC. Role of positron emission tomography-computed tomography in head and neck cancer. Ann Otol Rhinol Laryngol. 2011;120:593–602. 9. Chartrand-Lefebvre C, Howarth N, Grenier P, Keime F, Orcel B, Beigelman C. Association of small cell lung cancer and the antiHu paraneoplastic syndrome: radiographic and CT findings AJR Am J Roentgenol 1998;170:1513–157. 10. Dalmau J, Rosenfeld MR. Paraneoplastic syndromes of the CNS. Lancet Neurol. 2008;7:327–40. 11. Darnell RB, Posner JB. Paraneoplastic syndromes involving the nervous system. N Engl J Med. 2003;349:1543–54. 12. Frings M, Antoch G, Knorn P, Freudenberg L, Bier U, Timmann D, et al. Strategies in detection of the primary tumour in anti-Yo associated paraneoplastic cerebellar degeneration. J Neurol. 2005;252:197–201. 13. Graus F, Delattre JY, Antoine JC, Dalmau J, Giometto B, Grisold W, et al. Recommended diagnostic criteria for paraneoplastic neurological syndromes. J Neurol Neurosurg Psychiatry. 2004; 75:1135–40. 14. Graus F, Keime-Guibert F, Rene R, Benyahia B, Ribalta T, Ascaso C, et al. Anti-Hu-associated paraneoplastic encephalomyelitis: analysis of 200 patients. Brain. 2001;124:1138–48. 15. Hadjivassiliou M, Alder SJ, Van Beek EJ, Hanney MB, Lorenz E, Rao DG, et al. PET scan in clinically suspected paraneoplastic neurological syndromes: a 6-year prospective study in a regional neuroscience unit. Acta Neurol Scand. 2009;119:186–93.

Eur J Nucl Med Mol Imaging 16. Keime-Guibert F, Graus F, Fleury A, René R, Honnorat J, Broet P, et al. Treatment of paraneoplastic neurological syndromes with antineuronal antibodies (Anti-Hu, anti-Yo) with a combination of immunoglobulins, cyclophosphamide, and methylprednisolone. J Neurol Neurosurg Psychiatry. 2000;68:479–82. 17. Kumar A, Regmi SK, Dutta R, Kumar R, Gupta SD, Das P, et al. Characterization of thymic masses using (18)F-FDG PET-CT. Ann Nucl Med. 2009;23:569–77. 18. Linke R, Schroeder M, Helmberger T, Voltz R. Antibody-positive paraneoplastic neurologic syndromes: value of CT and PET for tumor diagnosis. Neurology. 2004;63:282–6. 19. Liu J, Xu Y, Wang J. Ultrasonography, computed tomography and magnetic resonance imaging for diagnosis of ovarian carcinoma. Eur J Radiol. 2007;62:328–34. 20. Liu RS, Yeh SH, Huang MH, Wang LS, Chu LS, Chang CP, et al. Use of fluorine-18 fluorodeoxyglucose positron emission tomography in the detection of thymoma: a preliminary report. Eur J Nucl Med. 1995;22:1402–7. 21. Matsuhisa A, Toriihara A, Kubota K, Makino T, Mizusawa H, Shibuya H, et al. Utility of F-18 FDG PET/CT in screening for paraneoplastic neurological syndromes. Clin Nucl Med. 2012;37:39–43. 22. Patel RR, Subramaniam RM, Mandrekar JN, Hammack JE, Lowe VJ, Jett JR, et al. Occult malignancy in patients with suspected paraneoplastic neurologic syndromes: value of positron emission tomography in diagnosis. Mayo Clin Proc 2008;83:917–22. 23. Rees JH, Hain SF, Johnson MR, Hughes RA, Costa DC, Ell PJ, et al. The role of [18F]fluoro-2-deoxyglucose-PET scanning in the diagnosis of paraneoplastic neurological disorders. Brain. 2001;124:2223–31.

24. Rojas I, Graus F, Keime-Guibert F, Reñé R, Delattre JY, Ramón JM, et al. Long-term clinical outcome of paraneoplastic cerebellar degeneration and anti-Yo antibodies. Neurology. 2000;55:713–5. 25. Rosenfeld MR, Dalmau J. Update on paraneoplastic neurologic disorders. Oncologist. 2010;15:603–17. 26. Rudnicki SA, Dalmau J. Paraneoplastic syndromes of the spinal cord, nerve, and muscle. Muscle Nerve. 2000;23:1800–18. 27. Titulaer MJ, Soffietti R, Dalmau J, Gilhus NE, Giometto B, Graus F, et al. Screening for tumours in paraneoplastic syndromes: report of an EFNS task force. Eur J Neurol. 2011;18:19–e13. 28. Titulaer MJ, Wirtz PW, Willems LN, van Kralingen KW, Smitt PA, Verschuuren JJ, et al. Screening for small-cell lung cancer: a follow-up study of patients with Lambert-Eaton myasthenic syndrome. J Clin Oncol. 2008;26:4276–81. 29. Tsujikawa T, Tsuchida T, Yoshida Y, Kurokawa T, Kiyono Y, Okazawa H, et al. Role of PET/CT in gynecological tumors based on the revised FIGO staging classification. Clin Nucl Med. 2011;36:e114–8. 30. Vedeler CA, Antoine JC, Giometto B, Graus F, Grisold W, Hart IK, et al. Management of paraneoplastic neurological syndromes: report of an EFNS Task Force. Eur J Neurol. 2006;13:682–90. 31. Voltz R. Paraneoplastic neurological syndromes: an update on diagnosis, pathogenesis, and therapy. Lancet Neurol. 2002; 1:294–305. 32. Younes-Mhenni S, Janier MF, Cinotti L, Antoine JC, Tronc F, Cottin V, et al. FDG-PET improves tumour detection in patients with paraneoplastic neurological syndromes. Brain. 2004; 127:2331–8.