Quantitative parameters correlated well with differentiation of squamous cell carcinoma at head and neck: a study of dynamic contrast-enhanced MRI

Abstract

Background

Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) is widely used for the diagnosis and prognostic assessment of head and neck squamous cell carcinoma (HNSCC). However, no research on grading HNSCC using DCE-MRI has been found. We hypothesize that DCE-MRI can grade the HNSCC non-invasively.

Purpose

To verify the hypothesis that DCE-MRI can grade the HNSCC non-invasively.

Material and Methods

Forty-two patients with histopathologically proved HNSCC from September 2013 to February 2016 were retrospectively analyzed. Chi-square test was used to compare patterns of time intensity curves (TICs) between well and poorly differentiated HNSCC. Two-sample t-test was performed to calculate the difference of volume transfer constant (K_trans), extravascular extracellular volume fraction (V_e), and initial area under the curve (iAUC) between groups. The diagnostic ability and cut-off value were assessed by receiver operator characteristic analysis.

Results

Most TICs of HNSCC are type III; no difference between well and poorly differentiated HNSCC has been found (P > 0.05). The value of K_trans, V_e, and iAUC for well and poorly differentiated HNSCC are (0.218 ± 0.048; 0.383 ± 0.074) min⁻¹, (0.605 ± 0.108; 0.712 ± 0.150), and (27.552 ± 6.238; 43.157 ± 9.148), respectively. K_trans, V_e, and iAUC are higher in poorly differentiated HNSCC, compared with well differentiated HNSCC (P < 0.001, 0.013, and < 0.001, respectively). K_trans has the greatest diagnostic significance with Youden’s index being 0.859 by cut-off value 0.270 min⁻¹. The diagnostic sensitivity and specificity were 95.0% and 90.9%, respectively.

Conclusion

The K_trans, V_e, and iAUC of HNSCC can be reliable quantitative parameters for evaluating well and poorly differentiated HNSCC where K_trans has the highest value.

Keywords

Dynamic contrast-enhanced magnetic resonance imaging head and neck squamous cell carcinoma grading differentiation

Introduction

Squamous cell carcinoma (SCC) is the most common primary malignant tumor in the head and neck, accounting for >3.2% of all malignancies (1). Of patients diagnosed with head and neck squamous cell carcinoma (HNSCC), ≤50% can survive for >5 years (2 –4). Though histologic grade is not included in the current staging criteria, histologic grade has a correlation with survival rate. An early study of 2007 cases of HNSCC by Roland et al. (5) advocated that poorly differentiated HNSCC correlated with a higher rate of nodal and distant metastasis and the recurrence rates increased 1.6–18% in poorly differentiated tumors compared to well differentiated HNSCC. The study claimed that the overall survival rate could decrease 6% in patients with poorly differentiated tumors. Piffko et al. (6) further demonstrated that histopathological malignancy was the most powerful prognostic indicator for cancer-specific mortality in patients with oral SCCs. Brandwein et al. (7) suggested that histologic risk assessment is a better prognosticator than margin status for predicting both local recurrence and overall survival of oral SCC. A more recent study by Thomas et al. (8) found that high histologic grade in early stage oral cavity cancer was associated with poorer survival and carried independent prognostic value in addition to tumor size, node status, and presence of distant metastasis (TNM) stage. Currently ultra-sonographic-guided fine-needle aspiration is often used for determining tumor differentiation. However, it is invasive, operator-dependent, and the results might be affected by the location of lesion and needle localization. Magnetic resonance imaging (MRI) provides more comprehensive information on the tumor including staging, margin, and lymph nodes status at the same time. Dynamic contrast-enhanced MRI (DCE-MRI) can provide hemodynamic information non-invasively (9 –11). It turns out that wash-in and wash-out patterns and quantitative parameters of DCE-MRI, such as volume transfer constant (K_trans), extravascular extracellular volume fraction (V_e), and initial area under the curve (iAUC), are useful in the analysis of the microcirculation of tissues (10,12–14). DCE-MRI is widely used for the diagnosis and prognostic assessment of head and neck cancers (15 –17), as well as for the evaluation of differentiation of glioma (18,19) and evaluation of microcirculation of extraocular muscle (13). However, to the best of our knowledge, few studies or not enough data in the literature about the grading of HNSCC by DCE-MRI parameters have been carried out.

We hypothesize that DCE-MRI can grade the HNSCC non-invasively. The purpose of our study is to compare: (i) the patterns of time intensity curves (TICs) between the poorly and well differentiated HNSCC; and (ii) the difference of quantitative parameters of DCE-MRI between the poorly and well differentiated HNSCC.

Material and Methods

Patients

This retrospective study was approved by the ethics committee of the hospital. The database was searched for all patients with microscopically proved HNSCC and all of the cases had available DCE-MRI in the hospital between September 2013 and February 2016. Grades I and II of the traditional grading system (20) were grouped as well-differentiated tumors, and Grades III and IV were grouped as poorly differentiated, based on a previous study by Piffko et al. (6). The patients were divided into two groups with well and poorly differentiated HNSCC according to the classification from the literature. Patients were excluded if: (i) the patient underwent therapy, operation, or biopsy before MRI; (ii) the time interval between MRI and histopathological examination was > 1 week; or (iii) image quality was heavily affected by motion artifact. Forty-two eligible cases were included in this study (30 men, 12 women). Basic information about the cases was listed in Suppl. Table 1.

Table 1.

MRI protocols.

	T1W imaging – 1	T2W imaging – 1	T1W imaging – 2	T2W imaging – 2	T2W imaging – FS	VIBE	TWIST
Plane	Axial	Axial	Coronal	Coronal	Axial/sagittal	Coronal	Coronal
Time to repeat (ms)	500	6000	600	4000	4000	5.29	4.9
Time to echo (ms)	8.2	94	8.2	84	84	1.81	2.0
Flip angle (°)	–	–	–	–	–	2/15	12
Slice thickness (mm)	2.5	2.5	3.0	3.0	2.5/3.0	3	3
Slice gap (mm)	0	0	0	0	0.5/0.6	0.6	0.6
FOV (mm)	170 × 240	170 × 240	240 × 200	240 × 200	240 × 200	240 × 200	240 × 200
Enhancement	No	No	No	No	No	No	Yes

MRI

MRI was performed on a 3.0-T unit (Trio Tim, Siemens AG, Erlangen, Germany) with an eight-channel head coil and eight-channel neck coil. Conventional sequences without enhancement were acquired before contrast-medium injection. Coronal T1 maps were acquired with volumetric interpolated breath-hold (VIBE) sequence. Coronal DCE images were acquired by time-resolved angiography with interleaved stochastic trajectories (TWIST) sequence. Fifty consecutive scans were acquired with one scan taking about 6.4 s. The total scan time was about 5 min 23 s. Magnevist (Bayer, Berlin, Germany) was injected with a dose of 0.1 mmol/kg of the patient’s weight at the beginning of the fourth scan. The injection rate was 2 mL/s, followed by 20-mL saline flushing. Image acquisition protocols were summarized in Table 1.

DCE-MRI process

All DCE-MR images of HNSCC were evaluated by a qualified radiologist with five years of experience. The DCE data were analyzed using Tissue 4D of Syngo, provided by Siemens Company. Regions of interest (ROI) were manually hand-drawn on the DCE images under the guidance of non-enhanced images. The whole volume of the tumors was included by placing freehand ROIs slice by slice, with an exclusion of areas of necrosis, hemorrhage, and vessels (Fig. 1). VOI analysis on Tissue 4D was selected to extracted whole tumors TIC and quantitative parameters.

Fig. 1.

Under the guidance of the coronal T2-weighted images, the whole volume of the tumor was included during the analysis by drawing slice by slice, with an exclusion of areas of necrosis and vessels.

Motion correction was performed before further analysis. The TICs of HNSCC exhibited three types: type I (Fig. 2a), TICs with gradual washing-in and without washing-out; type II (Fig. 2b), TICs with rapid and continuous washing-in and with little washing-out; type III (Fig. 2c), TICs with rapid early washing-in and obvious washing-out.

Fig. 2.

Types of TICs: (a) type I; (b) type II; and (c) type III.

The values of K_trans, V_e, and iAUC of the selected ROI were calculated using Tofts Model, based on population-averaged AIF (21).

Statistical analysis

All analysis was performed using the SPSS 17.0 software, with P < 0.05 being regarded as significant. All measurement parameters were expressed as mean ± standard deviation (SD). Kolomgorov–Smirnov test was used to check the normality of K_trans, V_e, and iAUC. Chi-square test was used to compare patterns of TICs between well and poorly differentiated HNSCC. Two-sample t-test was performed to calculate the difference of K_trans, V_e, and iAUC between two groups. Receiver operator characteristic (ROC) analysis was conducted to assess the diagnostic ability and the cut-off value with the highest Youden's index (Youden’s index = sensitivity + specificity – 1) was selected.

Results

Clinical data

A total of 42 patients were included in the study, while 22 patients (52.4%) had well differentiated HNSCC and 20 patients (47.6%) had poorly differentiated HNSCC. No significant differences of the age and gender distribution was noted between the groups with well and poorly differentiated HNSCC (P = 0.863 for age and P = 0.845 for gender).

Comparison of patterns of TICs

Over 66% of TICs of HNSCC were type III. Type I TIC was observed only in one patient with well differentiated HNSCC. No statistical difference was found between well and poorly differentiated HNSCC. The detailed information was shown in Table 2.

Table 2.

TICs of SCC.

Differentiation	Type I	Type II	Type III	P
Well	1 (4.5)	5 (22.7)	16 (72.8)	0.319
Poorly	0 (0.0)	2 (10.0)	18 (90.0)

Values are presented as n (%).

Comparison of quantitative parameters

K_trans was significantly higher in poorly differentiated tumors (0.383 ± 0.074 min⁻¹) than in well differentiated tumors (0.218 ± 0.048 min⁻¹) (T = –7.032, P < 0.001). V_e was significantly higher in poorly differentiated tumors (0.712 ± 0.150) than in well differentiated tumors (0.605 ± 0.108) (T = –2.938, P = 0.008). And iAUC was also significantly higher in poorly differentiated tumors (43.157 ± 9.148) than in well differentiated tumors (27.552 ± 6.238) (T = –17.015, P < 0.001). Detailed comparisons of value of K_trans, V_e, and iAUC between groups were presented in Table 3.

Table 3.

Values of K_trans, V_e, and iAUC of HNSCC, and ROC analysis results.

Parameters	Differentiation		P	Cut-off value	Accuracy (%)	Sensitivity (%)	Specificity (%)	NPV (%)	PPV (%)
Parameters	Well	Poorly	P	Cut-off value	Accuracy (%)	Sensitivity (%)	Specificity (%)	NPV (%)	PPV (%)
K_trans (min⁻¹)	0.218 ± 0.048	0.383 ± 0.074	<0.001*	0.270	92.9	95.0	90.9	95.2	90.5
V_e	0.605 ± 0.108	0.712 ± 0.150	0.013*	0.732	76.2	55.0	95.5	65.6	91.7
iAUC	27.552 ± 6.238	43.157 ± 9.148	<0.001*	31.552	88.1	95.0	81.8	94.7	82.6

*P < 0.05.

PPV, positive predictive value; NPV, negative predictive value.

The cut-off values and diagnostic criteria of parameters were present in Table 3. Area under the ROC curve (AUC) of K_trans, V_e, and iAUC were 0.964, 0.717, and 0.950, respectively (Fig. 3). K_trans was proved to be the parameter with greatest diagnostic ability with Youden’s index 0.895 by cut-off value 0.270 min⁻¹. The diagnostic accuracy, sensitivity, and specificity were 92.9%, 95.0%, and 90.9%, respectively.

Fig. 3.

Results of ROC analysis of quantitative parameters. AUCs of K_trans, V_e, and iAUC were 0.964, 0.717, and 0.950, respectively.

Discussion

There are two important findings reported in this paper. First, most TICs of HNSCC are type III, especially for poorly differentiated HNSCC. Second, the values of K_trans, V_e, and iAUC are higher in poorly differentiated HNSCC compared to those of well differentiated HNSCC. K_trans is the most sensitive parameter to discriminate well and poorly differentiated HNSCC.

In this study, the majority of TICs of HNSCC are type III. The result agrees with previous studies that revealed malignant head and neck tumors had faster wash-in and wash-out times (22,23), indicating microvascular circulation of malignant tumors were highly permeable. The permeability increases in accordance with the decrease of differentiation of HNSCC. However, the comparison of the TIC type between groups of cases stratified by HNSCC grade showed no statistical difference. The reason might be related to the locations of primary tumors. Individual conditions such as blood pressure and heart rate might influence the occurrences of wash-in and wash-out (24). Another possibility is that the classification of TICs is too generalized to provide more detailed information on microcirculation. Further studies are expected to find the exact cause for the discrepancy.

In the present study, we find that values of K_trans, V_e, and iAUC are statistically higher in poorly differentiated HNSCC compared to well differentiated HNSCC. To our knowledge, this is the only study on grading of HNSCC with DCE-MRI, whereas the DCE-MRI on grading of other solid tumors has been proven valuable by several researches (18,19,25). Choi et al. analyzed DCE-MRI of 33 patients diagnosed with pathologically confirmed gliomas (17). In their study, high-grade gliomas showed significant higher K_trans than low-grade gliomas. Jia et al. (19) got consistent results through research on 65 patients with oligodendrogliomas. They proved that mean K_trans was significantly lower for low-grade oligodendrogliomas than those for anaplastic oligodendrogliomas. Another study by Ma et al. (25) on breast cancer has also demonstrated that the K_trans value is a biomarker for grading breast cancer. It has been confirmed that the K_trans value from DCE-MRI could be used to estimate the permeability of pathological proliferated microvacuoles. Research by Lee et al. (26) has suggested that value of K_trans is positively correlated to expression of vascular endothelial growth factor in head and neck cancers. Similar results in HNSCC have been proved by Surov et al. in a recent study (22). Significant correlations were observed between V_e and mean vessel diameter and cell count. In addition, an inverse correlation between K_trans and K_i 67 was identified in primary HNSCC. Additionally, a study by Chawla et al. (27) proved that the K_trans parameter reflects a combination of tumor blood flow and microvascular permeability. It has been suggested that higher tumor blood flow results in better oxygenation and drug delivery to the target site, while hypoxia and decreased blood flow enhance chemo-resistance and impedes delivery of therapeutic agents. This indicates that higher K_trans values may predict prolonged survival, while lower K_trans values indicate a heterogeneous blood supply and poor survival (17,27). The result therefore may provide more information on the treatment plans and prognosis and is worth further study. The iAUC integrates the area under the kinetic curve during the first 90 s and is related to the increased permeability and angiogenesis. Endothelial cells of poorly differentiated HNSCC are often more immature than those of well differentiated HNSCC, lacking in pericyte and smooth muscle coverage, resulting in the increase in the permeability of new vessel in poorly differentiated HNSCC. The increased values of K_trans and iAUC of poorly differentiated HNSCC correspond to the histopathological changes. V_e, as a symbol of the extravascular volume fraction (11,28), is also analyzed to be significantly higher in poorly differentiated than that in well differentiated HNSCC. It is similar to the studies on glioma and breast cancer (18,19,25).

There are some limitations in this retrospective study that can be improved in further research. The sample size is relatively small; expending the sample size can improve the statistical significance of this work. Due to the small sample size, the location of HNSCC is not specified either. Therefore, a large cohort study is also expected to give diagnosis guidelines of the HNSCC according to the location. Additionally, the present study does not correlate the quantitative parameters obtained with the recurrence and survival rates of the patients. Although it proved to be robust on quantitative parameters based on whole tumor ROI analysis in head and neck cancer, this study lacks a second reader and statistical studying of inter-observer agreement for studied groups.

In conclusion, this study shows that the quantitative parameters are higher in poorly differentiated HNSCC. K_trans is of high accuracy for differentiation between well and poorly differentiated HNSCC. TICs play an unclear roll on grading HNSCC.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

References

Hernandez Brenda

Green Michael

Cassel Kevin

, et al. Preview of Hawaii cancer facts and figures 2010. Hawaii Med J 2010; 69:223–224.

Lefebvre

Chevalier

Luboinski

, et al. Larynx preservation in pyriform sinus cancer: preliminary results of a European Organization for Research and Treatment of Cancer phase III trial. J Natl Cancer Inst 1996; 88:890–899.

Wang

Hsu

Hsieh

, et al. Concurrent chemoradiotherapy using cisplatin, tegafur, and leucovorin for advanced squamous cell carcinoma of the hypopharynx and oropharynx. Biomed J 2014; 37:133–140.

Jafari

Najafi

Moradi

, et al. Delay in the diagnosis and treatment of oral cancer. J Dent (Shiraz) 2013; 14:146–150.

Roland

Caslin

Nash

, et al. Value of grading squamous cell carcinoma of the head and neck. Head Neck 1992; 14:224–229.

Piffkò

JBA

Ofner

Bryne

, et al. Prognostic value of histobiological factors (malignancy grading and AgNOR content) assessed at the invasive tumour front of oral squamous cell carcinomas. Br J Cancer 1997; 75:1543–1546.

Brandwein-Gensler

Teixeira

Lewis

, et al. Oral squamous cell carcinoma: histologic risk assessment, but not margin status, is strongly predictive of local disease-free and overall survival. Am J Surg Pathol 2005; 29:167–178.

Thomas

Stedman

Davies

Grade as a prognostic factor in oral squamous cell carcinoma: a population-based analysis of the data.

Laryngoscope 2014; 124:688–694.

Rodjan

de Graaf

van der Valk

, et al. Retinoblastoma: value of dynamic contrast-enhanced MR imaging and correlation with tumor angiogenesis. Am J Neuroradiol 2012; 33:2129–2135.

10.

Langer

van der Kwast

Evans

, et al. Prostate tissue composition and MR measurements: investigating the relationships between ADC, T2, K(trans), v(e), and corresponding histologic features. Radiology 2010; 255:485–494.

11.

Tofts

PS.

Modeling tracer kinetics in dynamic Gd-DTPA MR imaging.

J Magn Reson Imaging 1997; 7:91–101.

12.

Jansen

Carlson

, et al. Correlation of a priori DCE-MRI and (1)H-MRS data with molecular markers in neck nodal metastases: Initial analysis. Oral Oncol 2012; 48:717–722.

13.

Guo

Huo

Wang

, et al. Evaluating the microcirculation of normal extraocular muscles using quantitative dynamic contrast-enhanced magnetic resonance imaging. J Comput Assist Tomogr 2016; 40:419–423.

14.

Teo

Thng

Koh

, et al. Dynamic contrast-enhanced magnetic resonance imaging: applications in oncology. Clin Oncol (R Coll Radiols) 2014; 26:e9–20.

15.

Yuan

Yue

Tao

XF.

The diagnostic value of dynamic contrast-enhanced MRI for thyroid tumors.

Eur J Radiol 2012; 81:3313–3318.

16.

Jansen

Schöder

Lee

, et al. Tumor metabolism and perfusion in head and neck squamous cell carcinoma: pretreatment multimodality imaging with 1H magnetic resonance spectroscopy, dynamic contrast-enhanced MRI, and [18F]FDG-PET. Int J Radiat Oncol Biol Phys 2012; 82:299–307.

17.

Kim

Loevner

Quon

, et al. Prediction of response to chemoradiation therapy in squamous cell carcinomas of the head and neck using dynamic contrast-enhanced MR imaging. Am J Neuroradiol 2010; 31:262–268.

18.

Choi

Kim

Ahn

, et al. Glioma grading capability: comparisons among parameters from dynamic contrast-enhanced MRI and ADC value on DWI. Korean J Radiol 2013; 14:487–492.

19.

Jia

Geng

Liu

, et al. Low-grade and anaplastic oligodendrogliomas: differences in tumour microvascular permeability evaluated with dynamic contrast-enhanced magnetic resonance imaging. J Clin Neurosci 2013; 20:1110–1113.

20.

Edge

Byrd

Compton

, et al. AJCC Cancer Staging Manual. 7th ed. New York, NY: Springer-Verlag, 2010.

21.

Parker

Roberts

Macdonald

, et al. Experimentally-derived functional form for a population-averaged high-temporal-resolution arterial input function for dynamic contrast-enhanced MRI. Magn Reson Med 2006; 56:993–1000.

22.

Surov

Meyer

Gawlitza

, et al. Correlations between DCE MRI and histopathological parameters in head and neck squamous cell carcinoma. Transl Oncol 2017; 10:17–21.

23.

Guckel

Schnabel

Deimling

, et al. Dynamic snapshot gradient-echo imaging of head and neck malignancies: time dependency and quality of contrast-to-noise ratio. Magma 1996; 4:61–69.

24.

Giacomini

Jeffrey

Shin

LK.

Ultrasonographic evaluation of malignant and normal cervical lymph nodes.

Semin Ultrasound CT MR 2013; 34:236–247.

25.

Wang

Sun

, et al. Quantitative analysis of 3-Tesla magnetic resonance imaging in the differential diagnosis of breast lesions. Exp Ther Med 2015; 9:913–918.

26.

Lee

King

, et al. Dynamic contrast enhancement magnetic resonance imaging (DCE-MRI) for differential diagnosis in head and neck cancers. Eur J Radiol 2012; 81:784–788.

27.

Chawla

Kim

Loevner

, et al. Prediction of disease-free survival in patients with squamous cell carcinomas of the head and neck using dynamic contrast-enhanced MR imaging. Am J Neuroradiol 2011; 32:778–784.

28.

Tofts

Brix

Buckley

, et al. Estimating kinetic parameters from dynamic contrast-enhanced T(1)-weighted MRI of a diffusable tracer: standardized quantities and symbols. J Magn Reson Imaging 1999; 10:223–232.