T1-weighted dynamic contrast-enhanced brain magnetic resonance imaging: A preliminary study with low infusion rate in pediatric patients

Abstract

Background

The aim of this preliminary study is to evaluate the results of T1-weighted dynamic contrast-enhanced (DCE) magnetic resonance imaging (MRI) in pediatric patients at 1.5T, with a low peripheral intravenous gadoteric acid injection rate of 1 ml/s.

Materials and methods

Children with neurological symptoms were examined prospectively with conventional MRI and T1-weighted DCE MRI. An magnetic resonance perfusion analysis method was used to obtain time–concentration curves (persistent pattern, type-I; plateau pattern, type-II; washout pattern, type-III) and to calculate pharmacokinetic parameters. A total of two radiologists manually defined regions of interest (ROIs) in the part of the lesion exhibiting the greatest contrast enhancement and in the surrounding normal or contralateral tissue. Lesion/surrounding tissue or contralateral tissue pharmacokinetic parameter ratios were calculated. Tumors were categorized by grade (I–IV) using the World Health Organization (WHO) Grade. Mann–Whitney testing and receiver-operating characteristic (ROC) curves were performed.

Results

A total of nine boys and nine girls (mean age 10.5 years) were included. Lesions consisted of 10 brain tumors, 3 inflammatory lesions, 3 arteriovenous malformations and 2 strokes. We obtained analyzable concentration–time curves for all patients (6 type-I, 9 type-II, 3 type-III). K^trans between tumor tissue and surrounding or contralateral tissue was significantly different (p = 0.034). K^trans ratios were significantly different between grade I tumors and grade IV tumors (p = 0.027) and a K^trans ratio value superior to 0.63 appeared to be discriminant to determine a grade IV of malignancy.

Conclusions

Our results confirm the feasibility of pediatric T1-weighted DCE MRI at 1.5T with a low injection rate, which could be of great value in differentiating brain tumor grades.

Introduction

Dynamic contrast-enhanced (DCE) magnetic resonance imaging (MRI) is an emerging advanced imaging technique for pediatric patients, which overcomes the limitations of conventional MRI by providing supplementary physiological information such as tumor vascularity and hemodynamic characteristics of a cerebral neoplasm. In adults, MRI perfusion techniques such as DCE T1-weighted MRI have modified the therapeutic approach and follow-up assessment, particularly in neurologic oncology. Indeed, this perfusion technique, in association with conventional MRI, improves the accuracy, sensitivity, and specificity in differentiating tumors and non-neoplastic lesions in adult patients.^1,2 Its utility has been proven for differentiating glioblastoma, primary central nervous system lymphoma and brain metastatic tumor.³ DCE MRI could provide a consistent predictive factor (high Ve) of a poor evolution of preoperative dynamic contrast-enhanced MRI perfusion parameters for high grade glioma patients.⁴ It may also help identify patients with glioblastoma requiring close follow up after standard treatment.⁵ Translating such technique to pediatric populations could be useful. Important practical limitations make functional measurements more challenging in children, including logistical difficulties, movement during the procedure, particularly when the injection rate is high, and difficulty in obtaining peripheral intravenous access.⁶ With a less invasive approach, T1-weighted DCE MRI has two advantages: first, mathematical modeling of a bolus injection obviates the continuous infusion and plasma concentration monitoring and, second, the non-invasive measurement of tracer concentration. One difficulty is the need for a high perfusion bolus to observe either low signal in dynamic T2-weighted MRI or increased signal in T1. The high perfusion bolus is an important constraint in children, requiring a peripheral blood perfusion, a cumbersome procedure in daily practice. Software (MR Tissue 4D, Siemens, Germany) measures and converts the dynamic change of signal intensity in tissues. By measuring the signal–time curves, many pharmacokinetic parameters can be extracted and studied. For example, blood–brain barrier permeability and leakage space can be quantitatively measured by analyzing the dynamic curve. MRI cannot measure low normal transfer constants since gadolinium-diethylenetriamine pentaacetic acid enhancement is not observed in normal white matter. Recently, many reports based on T1-weighted DCE MRI have introduced various constants, (i.e. K^trans/min, volume transfer between extravascular, extracellular space and blood plasma, K_ep/min, rate transfer between extravascular, extracellular space and blood plasma, and V_e, extravascular, extracellular space volume per unit tissue volume). K^trans and V_e are pharmacokinetic parameters reflecting permeability. The transfer constant K^trans has several physiologic interpretations, depending on the balance between capillary permeability and blood flow in the tissue of interest. In high-permeability situations (where flux across the endothelium is flow limited), the transfer constant is equal to the blood plasma flow per unit volume of tissue. In the other limiting cases of low permeability, the transfer constant is equal to the permeability surface area product (between blood plasma and the extravascular extracellular space), per unit volume of tissue.^7,8 Vayapejam and colleagues have reported that pharmacokinetic parameters derived from DCE MRI can effectively discriminate between low and high grade pediatric brain tumors.⁹ Few measurements are available in high grade glioblastoma in children.¹⁰

In this report, we investigate the feasibility of T1-weighted DCE MRI in children at 1.5T MRI scanner with peripheral intravenous administration of gadoteric acid and gadolinium oxide at a low infusion rate (1 ml/s.)

Materials and methods

This prospective study was approved by our institutional review board. Parents provided their informed consent.

Patients

Between January 2016 and August 2016, we prospectively examined 18 children with conventional and T1-weighted DCE MRI in a pediatric university hospital. To be included, the patients had to be less than 16 years of age, have neurological symptoms, and required MRI as part of their routine care. All patients with abnormal MRI findings, and with the parents’ agreement, had brain DCE perfusion MRI. Our pediatric institution provides general and specialized neurological and neurosurgical care. Patients requiring general anesthesia for their scans and patients with renal function impairment, a known allergy to contrast material, or other contraindications to MRI were excluded. In children between 3 months and 5 years of age, MRI was performed with sedation, using hydroxyzine administered by mouth and intrarectal pentobarbital. Peripheral venous access required a 22-gauge catheter.

MRI protocol

All patients were scanned on a 1.5T scanner (Magnetom Aera, Siemens Healthcare, GmbH Germany) using a 20-channel head coil. Conventional MRI included the following sequences: Transverse T2-weighted turbo spin-echo; 3D T1-weighted magnetization-prepared rapid gradient-echo (MPRAGE); and transverse 3D fluid-attenuated inversion recovery (FLAIR).

DCE MRI was performed with 3D gradient-echo T1W (VIBE) after the intravenous administration of gadoteric acid and gadolinium oxide (0.2 ml/kg of body weight, Dotarem, Guerbet France; 0.1 mmol/kg). Contrast was administered using a power injector (Spectris MR injector; MedRad, Indianola, Pennsylvania, USA) through a peripherally-inserted intravenous cannula. The flowrate was 1 ml/sec for all patients. A 10-ml bolus injection of saline followed at the same injection rate. For each section, 56 images were acquired (at intervals of 14 sec). The parameters were as follows: time of repetition = 4.46 ms, time of echo = 1.72 ms, flip angle = 12°, matrix size = 166 × 256; slice thickness = 3 mm; field of view = 230 × 186 mm², voxel size = 0.9 × 0.9 × 3 mm³, pixel bandwidth = 300 Hz, and a total acquisition time of 2 minutes 53 seconds.

Image analysis

We processed the perfusion DCE MRI data by using the MRI perfusion analysis method (Syngo MR Tissue 4D, Siemens Healthcare GmbH, Erlangen, Germany), in which FLAIR or TSE T2-weighted images were used as a structural reference. On the basis of the two-compartment pharmacokinetic extended model proposed by Tofts and Kermode,^7,8 we used the perfusion analysis method to calculate pharmacokinetic parameters, including K^trans, K_ep and V_e, example in Figure 1. Deconvolution with the arterial input function (AIF) was performed in the pharmacokinetic model. In order to show the ideal relationship between the input function curve and the concentration–time curve, an appropriate AIF curve was chosen based on the lowest Chi-square value.

Figure 1.

Conventional MRI on image (a): post contrast 3D-T1 MPRAGE transverse-weighted images of an infratentorial medulloblastoma in an 8 year and 2 month-old boy. (b) Color map of K^trans, (c) K_ep and (d) V_e.

MPRAGE: magnetization-prepared rapid gradient-echo; MRI: magnetic resonance imaging

A total of two radiologists (senior resident, BBR; one faculty member with 4 years of experience in pediatric radiology; BM) manually defined two regions of interest (ROIs): one in the area of greatest enhancement in the lesion and the other one in the surrounding normal or contralateral non-pathologic tissue. In case of lack of consensus, both radiologists decided together to choose the most homogenous and well-limited ROI.

The concentration–time curve (CTC) obtained for each lesion and surrounding or contralateral tissue (axis coordinate, time; vertical coordinate, concentration) were analyzed qualitatively as three patterns based on the model described by Yuan and Al:¹¹ the persistent pattern with straight or curved line and continuous enhancement over the entire dynamic study was named type-I; the plateau pattern with a prominent increase slope and a final intensity 90–100% of peak grade was named type-II; and the washout pattern with a rapid increase slope and a final intensity <90% of peak grade was named type-III. In our study, the type-II pattern was divided into two different subtypes for clarity: The type-IIs in which the lesion’s CTC was higher than that of the contralateral non-pathological tissue; and the type-IIi in which the lesion’s CTC was lower than the non-pathological contralateral tissue CTC (Figure 2).

Figure 2.

Representation of the different concentration–time curve patterns.

Analysis of pharmacokinetic parameters

Our patients were classified into four groups based on the disease category: tumors, ischemic lesions, inflammatory lesions, and arteriovenous malformations. The tumor group was split into two subgroups depending on the 2016 World Health Organization (WHO) grade: low (I–II) or high (III–IV).

For each tumor, K^trans, K_ep and V_e ratios between the tumor and non-pathologic surrounding tissue were calculated. We compared separately each pharmacokinetic parameter measured in the tumor tissue and the non-pathologic surrounding or contralateral tissue, using a Mann–Whitney test and Student’s t-test.

For each individual pharmacokinetic parameter ratio, we compared among high grade and low grade tumor groups using a Mann–Whitney test and Student’s t-test. A significance level of <0.05 was applied. A receiver-operating characteristic (ROC) curve was determined. The area under curve (AUC) and accuracy depending on the cut-off threshold were estimated.

Results

Patients

A total of nine girls and nine boys (age range, 0.47–15.92 years; median age, 10.71 years; mean age, 10.55 years; see additional demographic and pathological details in Table 1) met the inclusion criteria. All patients were cooperative, tolerated the free-breathing imaging protocol without difficulty, and there were no movement-induced MRI artifacts.

Table 1.

Demographic data of children associated with lesion and concentration–time curves type.

Age	Sex	Lesion group	Pathology	Concentration–time curve type
3 months	G	High grade tumor	Rhabdoid tumor	I
8 years 2 months	B	High grade tumor	Medulloblastoma	III
10 years 10 months	B	High grade tumor	Medulloblastoma	I
9 years 11 months	B	High grade tumor	Medulloblastoma	IIs
15 years 11 months	G	Low grade tumor	Pilocytic astrocytoma	I
10 years 7 months	B	Low grade tumor	Pilocytic astrocytoma	Iii
7 years 9 months	G	High grade tumor	High grade glioma	III
11 years 4 months	B	Low grade tumor	Low grade glioma	III
8 years 11 months	B	High grade tumor	Glioma	IIs
15 years 3 months	G	Low grade tumor	DNET	IIs
15 years 4 months	G	Inflammatory disease	Multiple sclerosis (active lesion)	I
9 years 10 months	B	Inflammatory disease	Optic neuropathy	IIs
14 years 5 months	B	Inflammatory disease	Undetermined Inflammatory disease	IIs
15 years 2 months	B	Arteriovenous malformation	Cystic cavernoma	I
11 years 4 months	G	Arteriovenous malformation	Intracerebral arteriovenous malformation	IIs
8 years 4 months	B	Arteriovenous malformation	Subcutaneous arteriovenous malformation	IIs
2 years 1 month	G	Stroke	Old frontal stroke	III
13 years 11 months	B	Stroke	Recent vertebral dissection	Iii

B: boy; DNET: dysembryoplastic neuroepithelial tumour; G: girl

Lesions

Lesions consisted of 10 brain tumors, 3 inflammatory diseases, 3 arteriovenous malformations and 2 strokes. Details are reported in Table 1. The tumors were classified as low grade (I–II) in four patients and high grade (III–IV) in six patients, depending on the histological grade observed on biopsy.

Image analysis

We used the extended Tofts model with slow (n = 12) or intermediate (n = 6) AIF.

Concentration–time curve analysis

Analyzable concentration–time curves were available for all patients (6 type-I, 9 type-II, 3 type-III). Concentration–time curves are tabulated in Table 1. Among the tumors, there were 3 type-I (1 medulloblastoma, 1 pilocytic astrocytoma, 1 rhabdoid tumor); 1 type-IIi (1 pilocytic astrocytoma); 3 type-IIs (1 dysembryoplastic neuroepithelial tumour, 1 glioma, 1 medulloblastoma), and 3 type-III (2 gliomas and 1 medulloblastoma).

In the inflammatory group, there was 1 type-I (new lesion of multiple sclerosis), 1 type-IIi (1 optic neuropathy in pilocytic astrocytoma) and 1 type-IIs (undetermined inflammatory disease).

In the vascular malformation group, there was 1 type-I (cystic cavernoma), 1 type-IIi (intracerebral arteriovenous malformation) and 1 type-IIs (subcutaneous temporal arteriovenous malformation).

In the patients with an ischemic lesion, there was 1 type-IIi (a stroke due to a vertebral artery dissection) and 1 type-III (an old lesion due to a frontal stroke.)

Analysis of pharmacokinetics parameters

T1-weighted DCE MRI yielded a mean K^trans value of 0.0997 ± 0.19/min in the tumor group, 0.227 ± 0.16/min in the inflammatory disease group, 0.671 ± 0.861/min in the vascular malformation group, and 0.145 ± 0.096/min in the ischemic disease group (Table 2).

Table 2.

Mean values of pharmacokinetics parameters depending on pathology group

Disease Group	Mean Ktrans/min	Mean Kep/min	Mean Ve
Tumoral	0.100	4.584	0.033
Inflammatory	0.227	27.982	0.296
Arteriovenous malformation	0.671	3.300	0.432
Ischemic	0.145	5.300	0.034

For the tumors we studied, the T1-weighted DCE MRI indicated a mean K^trans value of 0.0477 ± 0.04/min for medulloblastoma, 0.24 ± 0.29/min for glioblastoma and 0.05 ± 0.017/min for pilocytic astrocytoma (Table 3).

Table 3.

Mean values of pharmacokinetic parameters depending on histological tumor group

Tumoral Type	Mean Ktrans/min	Mean Kep/min	Mean Ve
Medulloblastoma	0.048	2.040	0.050
Astrocytoma	0.050	4.790	0.030
Glioblastoma	0.240	7.245	0.027

K^trans was significantly different in tumor tissue compared with the non-pathologic surrounding or contralateral tissue (p = 0.034).

K^trans ratio was significantly different between grade IV and grade I (p = 0.027).

The ROC curves and accuracy curves for each pharmacokinetic parameter ratio in the group tumor are provided in Figure 3. The AUC of K^trans, K_ep and V_e ratios were 1, 0.75 and 0.77 respectively.

Figure 3.

Representation of the different ROC curves with accuracy analysis of K^trans (a and b) K_ep (c and d) and V_e (e and f) ratios.

ROC: receiver-operating characteristic

A K^trans ratio value superior to 0.63 appeared to be discriminant to determine a grade IV of malignancy. No K_ep or V_e ratio values in our sample appeared to be discriminant enough to determine the histological grade of malignancy.

K_ep

The T1-weighted DCE MRI yielded a mean K_ep value of 4.584 ± 4.22/min in the tumor group, 27.982 ± 20.58/min in the inflammatory disease group, 3.3 ± 3/min in the vascular malformation group, and 5.3 ± 1.264/min in the ischemic disease group (Table 2). In the tumor group, the T1-weighted DCE MRI indicated a mean K_ep value of 2.04 ± 0.46/min for medulloblastoma, 7.245 ± 5.06/min for glioblastoma and 4.79 ± 3.57/min for pilocytic astrocytoma.

In patients with tumors, the K_ep was not significantly different between the tumor tissue and the non-pathologic surrounding or contralateral tissue (p = 0.53)

The K_ep ratio was not significantly different between grade IV and grade I (p = 0.17).

V_e

The T1-weighted DCE MRI indicated mean V_e value of 0.033 ± 0.028 in the tumor group, 0.296 ± 0.41 in the inflammatory disease group, 0.432 ± 0.0265 in the vascular malformation group and 0.0335 ± 0.0265 in the ischemic disease group (Table 2).

In the tumor group, the T1-weighted DCE MRI yielded a mean V_e value of 0.05 ± 0.031 for medulloblastoma, 0.027 ± 0.018 for glioblastoma and 0.0295 ± 0.0255 for pilocytic astrocytoma (Table 3).

The V_e was significantly different in the tumor group, when tumor tissue was compared with non-pathologic surrounding tissue (p = 0.009).

The V_e ratio was not significantly different between grade IV and grade I (p = 0.26).

Discussion

This preliminary study confirmed the feasibility of T1-weighted DCE MRI with an intravenous injection of gadoteric acid and gadolinium oxide at a rate of 1 ml/s in children at 1.5T. CTCs provided valuable perfusion information, particularly in arteriovenous malformation, and useful kinetic enhancement data. Our results indicate that the characteristics of the CTC could help differentiate tumor types, as demonstrated in adults.^12,13 All malignant tumors showed a plateau or washout pattern, but no persistent pattern, suggesting the link between malignancy and microvascular circulation, as shown by Zahra and colleagues.¹⁴

T1-weighted DCE MRI could help to determine the tumor grade, allowing new perspectives in pediatric clinical care. The relationship between pharmacokinetic parameters obtained by DCE perfusion imaging and tumor grade has been investigated in adults¹⁵ but not in the pediatric population. The study of Vajapeyam and colleagues had been conducted with higher infusion rates (2 ml/s) at 3T and showed that pharmacokinetic parameters correlated with tumor grade.⁹ Very few data have been reported on high grade glioblastoma in children, with pharmacokinetic parameters.¹⁰ We could suggest a cut-off K^trans ratio value for distinguishing low from high grade pediatric malignancy brain tumors. Our pharmacokinetic parameters were close to but not identical with those previously reported. The difference may be explained by the number of acquisition parameters (lower magnetic field and infusion rate) and by the underlying model provided by the software. For example, it is known that DCE MRI-derived pharmacokinetic parameters are heavily dependent on the model and input parameters used^16–18 and are difficult to standardize. To minimize these differences, we suggest using a ratio between K^trans in the tumor and K^trans in the surrounding or contralateral healthy brain tissue, allowing better differentiation of tumor grade. Ho and colleagues^19,20 have found a significant positive correlation with tumor grade by using relative blood flow data from dynamic susceptibility contrast perfusion in pediatric patients but also an important overlap between low and high grade. Previous studies with dynamic enhanced imaging or arterial spin labelling did not help distinguish between histological tumor types.^20,21 Our findings suggest that pharmacokinetic metrics such as K^trans and K_ep are distinct among various types of diseases, particularly between tumor types.

Our investigation is limited by the small sample size and by the heterogeneity of the disease groups and tumor types. However, our preliminary results demonstrate the interest of DCE MRI in discriminating among a variety of lesions, suggesting that such an approach might be broadly applicable. The infusion rate we utilized of 1 ml/s appears to be one of the lowest reported and may not be sufficiently high to obtain values of pharmacokinetic parameters significantly different to allow discrimination among various tumor types.

Another possible limitation was the measurement method we employed. The ROI was manually drawn in a limited part of the tumor. However, recent studies have demonstrated that ROI limited to the area of greatest early enhancement in the tumor reflects hemodynamic alteration of the tumor better than an ROI that includes the entire tumor.^22,23 The most malignant shares of the tumor had been assessed. A multiple ROI approach that allowed an assessment of the heterogeneity of the tumors could have been done.

Pharmacokinetic parameters seem to provide additional information useful in characterizing neuropathology, particularly in neoplastic lesions. However, further studies are needed to confirm and increase accuracy in interpreting the results.

Conclusion

T1-weighted DCE MRI can be performed in the pediatric population with an intravenous injection rate of 1 ml/sec at 1.5T. Quantitative CTCs can be obtained that help to differentiate tumor grades. In the patients with cerebral neoplasm in our study, K^trans was significantly different in tumor tissue compared with non-pathologic adjacent or contralateral tissue, and the K^trans ratio was significantly different between grade IV and grade I.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Conflict of interest

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