Patients’ characteristics
The socio-economical and clinical characteristics of the 117 studied patients are presented in Table 1. The mean age of patients was 43.2 with extreme ages of 12 and 80 years old. NPC was more prevalent in men (64.1%) than women (35.9%), with a sex ratio of 1.7. The main symptoms reported by patients at disease onset showed that most patients presented with lymph node syndromes (82.1%) and headache (79.5%) as early-onset symptoms. The other common presenting symptoms were otologic issues and nasal obstruction reported in 70.9% (83/117) and 61.5% (72/117) of patients, respectively. The non-keratinizing undifferentiated carcinoma prevails and was found in 95.7% of patients (112/117). According to the 7th edition of the AJCC/UICC staging system, 23.9% of patients with NPC were classified in group III (28/117) and 65.0% in group IV (76/117). Imaging outcomes showed that most patients had tumor stages T3 (38/117) and T4 (50/117) and nodal stages N1 (30/117) and N2 (52/117). Furthermore, 31.6% (37/117) of patients have already developed metastasis at time of diagnostic. Evaluation of the local extension of nasopharyngeal tumors by HN-MRI showed that out of the 93 patients, 12 patients (12.9%) presented a limited tumor to the nasopharynx area and 81 patients (87.1%) have parapharyngeal tumor extension. Moreover, the tumor involved the bony structure of skull base in 53 patients (57.0%), and neurologic structures in 35 patients (37.6%).
Table 1
Patient’s characteristics
| Characteristics | Number of cases | % |
Gender | Female | 42/117 | 35.9 |
Male | 75/117 | 64.1 |
Age | ≤ 32 years old | 32/117 | 27.4 |
< 32 years old | 85/117 | 72.6 |
Smoking status | No | 100/117 | 85.5 |
Yes | 17/117 | 14.5 |
Drinking status | No | 100/117 | 85.5 |
Yes | 17/117 | 14.5 |
Clinical presentations | Headache | 93/117 | 79.5 |
Lymph node syndromes | 96/117 | 82.1 |
Unilateral | 53/117 | 45.3 |
Bilateral | 43/117 | 36.8 |
Otologic issues | 83/117 | 70.9 |
Nasal obstruction | 72/117 | 61.5 |
Epistaxis | 63/117 | 53.8 |
Neck swelling | 28/117 | 23.9 |
Trismus | 34/117 | 29.1 |
Neurologic Syndrome | 12/117 | 10.3 |
Histopathology | Non-keratinizing undifferentiated carcinoma | 112/117 | 95.7 |
Non-keratinizing differentiated carcinoma | 2/117 | 1.7 |
Keratinizing squamous cell carcinoma | 1/117 | 0.9 |
Others | 2/117 | 1.7 |
Tumor stage | T1 | 12/117 | 10.3 |
T2 | 17/117 | 14.5 |
T3 | 38/117 | 32.5 |
T4 | 50/117 | 43.7 |
Nodal stage | N0 | 10/117 | 8.5 |
N1 | 30/117 | 25.6 |
N2 | 52/117 | 44.4 |
N3 | 25/117 | 21.4 |
Metastatic status | M0 | 80/117 | 68.4 |
M1 | 37/117 | 31.6 |
Stage of the disease (NPC stage) | I | 2/117 | 1.7 |
II | 11/117 | 9.4 |
III | 28/117 | 23.9 |
IV | 76/117 | 65.0 |
Local extension of nasopharyngeal tumors by HN-MRI | Limited tumor to the nasopharynx area | 12/93 | 12.9 |
Parapharyngeal involvement | 81/93 | 87.1 |
Skull base bone invasion | 53/93 | 57.0 |
| Neurologic involvement | 35/93 | 37.6 |
Concordance between HN-MRI and [18F] FDG PET/CT in T and N staging
Among the 117 included patients, 77 patients underwent both HN-MRI and 18F-FDG-PET/CT, and were therefore used to evaluate imaging results in order to assess these two techniques in T and N staging accordingly. Tumor staging by HN-MRI showed that 34 patients were classified T4 (44.2%), 25 were classified T3 (32.5%), 12 were classified T2 (15.6%) and 6 were classified T1 (7.8%). T disease staging using the [18F] FDG PET/CT was as follows: 25 patients were classified T4 (33.8%), 29 were T3 (37.7%), 15 were T2 (18.2%) and 8 were classified T1 (10.4%). The [18F] FDG PET/CT correctly classified 59 patients (76%), underestimated 9 patients (11.6%) and overestimated 9 patients (11.6%), which makes a total of 18 misclassified patients (23%). Overall, a moderate concordance was observed between T-TEP and T-MRI categories with a Cohen kappa coefficient at 0.45. Regarding nodal staging, HN-MRI and [18F] FDG PET/CT showed a perfect concordance for N2 class, with 35 patients staged N2 (45.5%), while HN-MRI and [18F] FDG PET/CT did not detect nodal invasion in 10 (13%) and 8 patients (10.4%), respectively. Moreover, HN-MRI detected unilateral nodal involvement (N1) in 18 patients (23.4%) and N3 in 14 patients (18.2%), while [18F] FDG PET/CT detect N1 nodal involvement in 21 patients (27.3%) and N3 in 13 patients (16.9%). This difference between [18F] FDG PET/CT and HN-MRI in N staging was significant with a mediocre degree of concordance (k = 0.3). HN-MRI correctly classified nodal involvement of 71 patients (92%), underestimated nodal involvement of three patients (3.8%), and overestimated nodal involvement of three patients (3.8%), which makes a total of six (7.7 %) misclassified patients (Table 2).
Table 2
Concordance between HN-MRI and [18F] FDG PET/CT in T and N staging
Tumor stage | T MRI | T TEP | | | Nodal stage | N MRI | N TEP | |
T1 | 6 (7.8%) | 8 (10.4%) | | | N0 | 10 (13.0%) | 8 (10.4%) | |
T2 | 12 (15.6%) | 15 (18.2%) | K = 0.459 p = 0.00 | | N1 | 18 (23.4%) | 21 (27.3%) | K = 0,30 p = 0.00 |
T3 | 25 (32.5%) | 29 (37.7%) | | | N2 | 35 (45.5%) | 35 (45.5%) | |
T4 | 34 (44.2%) | 25 (33.8%) | | | N3 | 14 (18.2%) | 13 (16.9%) | |
Association analysis between metabolic [18F] FDG PET/CT parameters, TNM classes and NPC stage
Overall, the mean T-SUV max of the studied patients was 11.8 (3.30–36.10), the mean N-SUV max was 9.2 (1.50–19.8) and the mean M-SUV max was 7.4 (1.60–21.50). The NTR was also calculated, and showed a mean of 0.82 (0.16 to 1.9). The best cutoffs values for T-SUV max, N-SUV max and NTR were 10.5, 7.9 and 0.82, respectively (Fig. 1). Association analysis between metabolic [18F] FDG PET/CT parameters, TNM classes and NPC stage was further performed (Table 3). Our data showed that N-SUV max values were significantly higher in patients with advanced nodal involvement, with a mean of 7.4, 9.7 and 11 for patients with N1, N2, and N3 nodal classes, respectively (p = 0.01). However, no significant difference was observed between N-SUV max, tumor size, metastatic status and NPC stage (p > 0.05). Similarly, no significant associations were found between [18F] FDG PET/CT metabolic parameters (T-SUV max, M-SUV max, NTR), TNM classes and NPC stage (p > 0.05).
Table 3
Association analysis between metabolic [18F] FDG PET/CT parameters, TNM classes and NPC stage
| | T-SUV max | | | N-SUV max | | | NTR | | | M-SUV max | |
| N | Mean (range) | p | N | Mean (range) | p | N | Mean (range) | p | N | Mean (range) | p |
Tumor extension |
T1 | 11 | 10.0(4.2–14.3) | 0.03 | 9 | 8.6 (4.3–11.7) | 0.6 | 9 | 0.96 (0.5–1.9) | 0.3 | 4 | 6.7 (3.0-12.2) | 0.1 |
T2 | 15 | 9.4 (5.0-18.1) | 14 | 8.6 (2.9–15.2) | 14 | 0.90 (0.3–1.6) | 3 | 6.5 (4.9–8.6) |
T3 | 32 | 11.5 (3.3–25.4) | 29 | 9.1 (1.5–17.8) | 29 | 0.81 (0.1–1.9) | 10 | 5.2 (1.6–13.4) |
T4 | 41 | 13.5 (5.8–36.1) | 39 | 10.0 (3.4–32.0) | 39 | 0.76 (0.2–1.7) | 14 | 9.4 (3.8–21.5) |
Nodal involvement |
N0 | 8 | 10.9(5.8–13.5) | 0.6 | - | - | 0.01 | - | - | 0.1 | - | - | 0.2 |
N1 | 26 | 11.0 (3.3–25.4) | 26 | 7.4 (1.5–16.5) | 26 | 0.75 (0.16–1.91) | 4 | 5.1(1.6–7.8) |
N2 | 41 | 12.3(4.3–19.8) | 41 | 9.7 (1.6–15.2) | 41 | 0.82(0.26–1.97) | 13 | 6.8(4.9–12.6) |
N3 | 24 | 12.5 (5.1–36.1) | 24 | 11.0 (4.2–32.0) | 24 | 0.91 (0.52–1.47) | 14 | 8.6(1.6–21.5) |
Metastatic status |
M0 | 63 | 12.1 (4.2–36.1) | 0.5 | 55 | 9.1 (1.6–32.0) | 0.5 | 55 | 0.78(0.26–1.91) | 0.1 | | | |
M1 | 36 | 11.4 (3.3–20.3) | 36 | 9.7 (1.5–19.8) | 36 | 0.88 (0.16–1.97) | | |
Stage of the disease (NPC stage) |
I | 2 | 11.1(8.7–13.5) | 0.2 | - | - | 0.4 | - | - | 0.1 | | | |
II | 9 | 8.3 (4.2–13.2) | 8 | 7.4 (2.9–14.2) | 8 | 1.06 (0.32–1.91) | | | |
III | 23 | 12.3 (4.3–25.4) | 20 | 9.7 (1.6–16.5) | 20 | 0.78(0.26–1.18) | | | |
IV | 65 | 12.2 (3.3–36.1) | 63 | 9.4 (1.5–32.0) | 63 | 0.81 (0.16–1.97) | | |
Radiological prognostic factors related to OS of NPC patients
After a mean follow-up time of 37 months (20–54 mo), 66.3% of patients were alive with mean 4-year OS of 22.7 months. In the present study, the evaluation of radiological and metabolic prognostic factors influencing OS was performed. Our results showed that OS was better in patients with N-SUV max < 7.9 versus ≥ 7.9 (mean 38.4 vs. 31.4 mo, respectively; HR, 0.70; 95% CI, 0.48–1.01; p = 0.05) and NTR < 0.81 versus ≥ 0.81 (mean 40.4 vs. 26.5 mo, respectively; HR, 0.55; 95% CI, 0.38–0.78; p = 0.001); suggesting higher N-SUV max and NTR as an independent poor prognostic factors for NPC. However, no association between T-SUV max, parapharyngeal tumor extension, skull base bone tumor invasion, neurologic involvement and OS was observed among NPC patients (p > 0.05) (Table 4).
Table 4
Radiological factors related to OS of NPC patients
| OS |
| | 4-year rate (%) | Mean (months) | | Univariate analysis | Multivariate analysis |
HR (95% CI) | p | HR (95% CI) | p |
T-SUV max | 99 | | | | | | | |
< 10.5 | 45 | 59,8% | 33.8 | | 1.11 (0.58–2.15) | 0.74 | - | - |
≥ 10.5 | 54 | 61.5% | 37.1 |
N-SUV max | 91 | | | | | | | |
< 7.9 | 35 | 70.0% | 38.4 | | 0.70 (0.48–1.01) | 0.05 | 0.39 (0.17–0.89) | 0.02 |
≥ 7.9 | 56 | 48.9% | 31.4 |
NTR | 91 | | | | | | | |
< 0.81 | 47 | 73.8% | 40.4 | | 0.55 (0.38–0.78) | 0.001 | 0.41 (0.18–0.94) | 0.03 |
≥ 0.81 | 44 | 38.8% | 26.5 |
Parapharyngeal involvement | 93 | | | | | | | |
No | 12 | 82.5% | 43.8% | | 0.76 (0.36–1.56) | 0.4 | - | - |
Yes | 81 | 67.0% | 40.0% |
Skull base bone invasion | 93 | | | | | | | |
No | 40 | 71.4% | 42.3 | | 0.88 (0.59–1.30) | 0.5 | - | - |
Yes | 53 | 67.1% | 38.9 |
Neurologic involvement | 93 | | | | | | | |
No | 58 | 67.5% | 40.9 | | 0.96 (0.65–1.43) | 0.8 | - | - |
Yes | 35 | 70.6% | 39.5 |
Radiological prognostic factors related to LRRFS of NPC patients
Univariate analysis showed that LRRFS was not significantly affected by T-SUV max, N-SUV max, NTR and neurologic involvement with HRs of 0.97 (95% CI: 0.42–2.26; p = 0.9), 0.45 (95% CI: 0.17–1.18; p = 0.1), and 0.71 (95% CI: 0.30–1.70; p = 0.4), respectively. Similarly, our results revealed that patients without parapharyngeal tumor extension, skull base bone tumor invasion and neurologic involvement have better LRRFS with a trend of significance, compared to those with parapharyngeal tumor extension, bone tumor invasion and neurologic involvement: mean 43.4 vs. 40.5 mo, and an HR of 0.60 (95% CI, 0.13–2.61; p = 0.4) ,mean 44.6 vs. 38.6 mo and an HR of 0.51 (95% CI, 0.20–1.28; p = 0.1) and mean 41.4 vs. 40.6 mo and an HR of 0.95 (95% CI: 0.39–2.31; p = 0.9), respectively (Table 5).
Table 5
Radiological factors related to LRRFS of NPC patients
| LRRFS |
| N | 4-year rate (%) | Mean (months) | | Univariate Analysis |
HR(95% CI) | p |
T-SUV max | 99 | | | | | | |
< 10.5 | 45 | 67.4% | 39.1 | | 0.97 (0.42–2.26) | 0.9 |
≥ 10.5 | 54 | 54.8% | 40.7 |
N-SUV max | 91 | | | | | | |
< 7.9 | 35 | 75.9% | 42.7 | | 0.45 (0.17–1.18) | 0.1 |
≥7.9 | 56 | 51.0% | 37.3 |
NTR | 91 | | | | | | |
< 0.81 | 47 | 69.0% | 39.8 | | 0.71 (0.30–1.70) | 0.4 |
≥ 0.81 | 44 | 51.1% | 38.6 |
Parapharyngeal involvement | 93 | | | | | | |
No | 12 | 76.2% | 43.4 | | 0.60 (0.13–2.61) | 0.4 |
Yes | 81 | 56.5% | 40.5 |
Skull base bone invasion | 93 | | | | | | |
No | 40 | 68.5% | 44.6 | | 0.51 (0.20–1.28) | 0.1 |
Yes | 53 | 59.8% | 38.6 | |
Neurologic involvement | 93 | | | | | | |
No | 58 | 68.3% | 41.4 | | 0.95 (0.39–2.31) | 0.9 |
Yes | 35 | 47.1% | 40.6 |
Radiological prognostic factors related to PFS of NPC patients
Univariate analysis showed that PFS was significantly shorter among patients with high N-SUVmax (mean 31.0 vs. 21.4 mo; HR, 0.52; 95% CI, 0.29–0.93; p = 0.02) and high NTR (mean 30.4 vs. 18.7 mo; HR, 0.46; 95% CI, 0.27–0.79; p = 0.003). T- SUVmax, parapharyngeal tumor extension and neurologic involvement were not found to represent a significant prognostic factor for PFS in NPC (p = 0.7, p = 0.3and p = 0.2, respectively). Of note, a trend of decreased PFS time and rates was observed in patients with skull base bone tumor invasion compared to patients with limited tumor (38.0% vs. 51.7%; 27.6 vs 36.3 mo; HR, 0.54; 95% CI, 0.29–1.02; p = 0.05) (Table 6).
Table 6
Radiological factors related to PFS of NPC patients
| PFS |
| N | 4-year rate (%) | Mean (months) | | Univariate analysis | Multivariate analysis |
HR (95% CI) | p | HR (95% CI) | p |
T-SUV max | 99 | | | | | | | |
< 10.5 | 45 | 39.8% | 26.6 | 0.93 (0.55–1.57) | 0.7 | - | - |
≥ 10.5 | 54 | 29.9% | 26.4 |
N-SUV max | 91 | | | | | | | |
< 7.9 | 35 | 47.8% | 31.0 | 0.52 (0.29–0.93) | 0.02 | 0.46 (0.23–0.90) | 0.04 |
≥ 7.9 | 56 | 23.2% | 21.4 |
NTR | 91 | | | | | | | |
< 0.81 | 47 | 46.0% | 30.4 | 0.46 (0.27–0.79) | 0.003 | 0.50 (0.28–0.89) | 0.01 |
≥ 0.81 | 44 | 18.1% | 18.7 |
Parapharyngeal involvement | 93 | | | | | | | |
No | 12 | 63.5% | 36.6 | 0.60 (0.21–1.68) | 0.3 | - | - |
Yes | 81 | 37.6% | 30.2 |
Skull base bone invasion | 93 | | | | | | | |
No | 40 | 51.7% | 36.3 | 0.54 (0.29–1.02) | 0.05 | 0.18 (0.07–0.50) | 0.001 |
Yes | 53 | 38.0% | 27.6 |
Neurologic involvement | 93 | | | | | | | |
No | 58 | 49.3% | 33.2 | 0.69 (0.38–1.25) | 0.2 | - | - |
Yes | 35 | 29.3% | 27.6 |
Multivariate Analysis for PFS and OS
Only significant prognostic factors identified in univariate analysis for OS and PFS were assessed in the multivariate analysis (Tables 4 and 5). Among all the radiological and metabolic factors studied, we identified N-SUV max and NTR as independent prognostic factors for both OS and PFS (p < 0.05), and skull base bone invasion as independent prognostic factor for PFS in patients with NPC (p = 0.001).