Abstract
Purpose
The purpose of this study is to evaluate the accuracy of apparent diffusion coefficient magnetic resonance imaging in grading tumor aggressiveness using histogram apparent diffusion coefficient values.
Materials and methods
Eighteen patients with surgically proved pituitary macroadenomas were included in this study. Diffusion-weighted imaging with single-shot echo-planar sequence at 3-T with a 32-channel head coil was performed with b values of 0 and 1000 s/mm2. Calculated apparent diffusion coefficient maps were generated, and a 3-D volume of interest was placed on the tumor while superimposing contrast-enhanced magnetic resonance images. All apparent diffusion coefficient values within the volume of interest were used to compute the average apparent diffusion coefficient of the tumor. The apparent diffusion coefficient values were binned to construct the apparent diffusion coefficient histogram. Using the histogram, the mean, percentiles, skewness, and kurtosis of the apparent diffusion coefficient of the entire tumor were computed. Apparent diffusion coefficient histogram parameters were compared with the MIB-1 index, invasiveness, and recurrence for grading tumor aggressiveness of pituitary adenomas.
Results
The skewness of the apparent diffusion coefficient histogram only showed significant differences among MIB-1 indices (p = 0.030). All apparent diffusion coefficient histogram parameters showed no significant differences between negative and positive invasion. The skewness and kurtosis of the apparent diffusion coefficient histogram showed significant differences between positive and negative recurrence (skewness p = 0.011, kurtosis p = 0.011). Receiver-operating characteristics analysis between positive and negative recurrence showed that both skewness and kurtosis of the apparent diffusion coefficient achieved area under the curve at 0.967.
Conclusion
Skewness and kurtosis of the apparent diffusion coefficient histogram were the predictive parameters for assessing tumor proliferative potential and recurrence of pituitary adenomas.
Keywords: Apparent diffusion coefficient histogram, aggressiveness, pituitary adenoma
Introduction
Pituitary adenomas often accompany invasion and recurrence.1–3 In the 2017 World Health Organization (WHO) classification, tumor subtyping, tumor proliferative potential (mitotic account, MIB-1 index), and clinical parameters (tumor invasion and recurrence) are recommended for consideration of aggressive adenomas.4 MIB-1 is a clone name of a monoclinal antibody of the Ki-67 antigen. The Ki-67 antigen is a protein present in mitosis of the nuclei of cells. The MIB-1 index is expressed as a percentage of Ki-67 antigen positive nuclei among total nuclei.5
From a clinical point of view, preoperative evaluation of the aggressiveness of the tumor is important. Up to now, imaging criteria for tumor aggressiveness of pituitary adenomas have not been established. Diffusion-weighted imaging (DWI) provides information on the inhibitory effect of cell membranes based on the mobility of water molecules in tissues. Malignant lesions which have densely packed tumor cells with a high nuclear-to-cytoplasmic ratio can reduce water molecule motion. The apparent diffusion coefficient (ADC) values of malignant tumors show low values.6–8 Therefore, DWI is useful for differentiating malignant from benign tumors based on their different cellular densities. ADC histograms provide quantitative information about the tissue characteristics and heterogeneity of the entire tumor.9 There have been several reports on the utility of the histogram analysis of ADC values for distinguishing malignant and benign tumors.10–12
Single-shot echo-planar DWI is the most commonly used method, but single-shot echo-planar DWI sequences at the region of skull base are prone to susceptibility artifacts. On the other hand, the image quality of DWI can be improved by using the parallel imaging technique.13 Parallel imaging is a technique that involves the spatial sensitivity information inherent in an array of multiple-receiver surface coils in order to reduce time-consuming spatial encoding.13 DWI with parallel imaging has the advantage of reduced scan time and improved spatial resolution.
The purpose of this study is to evaluate the utility of histogram analysis of ADC values to evaluate the aggressiveness of pituitary adenomas using the parallel imaging technique.
Materials and methods
Patients
The institutional ethics committee approved this retrospective study and waived the requirement for informed consent (approval number I-155). Between 1 January 2012–31 December 2015, 18 patients (three men and 15 women; mean 55.2 years, range 37–77 years) with pituitary macroadenomas in the initial surgical cases underwent DWI, and corresponding ADC maps were obtained.
In order to avoid the partial volume effect of uniformity with surrounding tissue in cases when the tumor size was small, we evaluated pituitary adenomas that were more than 10 mm in size. Among these patients, five cases with cystic or hemorrhagic changes of size identifiable in magnetic resonance (MR) images were excluded from the analysis. In the remaining 13 cases, pathological confirmations including immunohistochemical examination were obtained using the resected specimen obtained by transsphenoidal surgery.
Table 1 shows the patient profile (age and sex), and tumor size, tumor type, MIB-1 proliferative index, and Knosp classification. Knosp classification (grades 0–4, grade 0: tumor does not extend the medial carotid line, grade 1: tumor crosses the medial line, but does not extend past the median line, grade 2: tumor extend beyond the medial line, but does not extend beyond the lateral line, grade 3: tumor extends beyond the lateral line, grade 4: intracavernous carotid artery covered with tumor) reflects the degree of invasion of the cavernous sinus using MR images.14
Table 1.
Pituitary macroadenoma patient and tumor characteristics.
| Case | Age (years) | Sex | Tumor size (mm) | Tumor secreting type | Knosp classification | MIB-1 index |
|---|---|---|---|---|---|---|
| 1 | 40 | Female | 15.6 | Non | 0 | 2% |
| 2 | 72 | Male | 25.8 | Non | 4 | <1% |
| 3 | 77 | Female | 18.9 | Non | 2 | <1% |
| 4 | 69 | Female | 23.5 | Non | 2 | 1–2% |
| 5 | 69 | Female | 14.0 | TSH | 2 | <1% |
| 6 | 37 | Female | 16.3 | Non | 1 | 2.7% |
| 7 | 37 | Female | 15.2 | GH | 3 | <1% |
| 8 | 44 | Female | 22.6 | Non | 4 | <1% |
| 9 | 63 | Female | 18.0 | Non | 0 | 1% |
| 10 | 59 | Female | 22.8 | Non | 1 | 1% |
| 11 | 40 | Male | 20.0 | Non | 3 | 4.2% |
| 12 | 42 | Female | 24.0 | PRL | 3 | 7.6% |
| 13 | 55 | Female | 24.0 | Non | 1 | 3% |
GH: growth hormone; Non: non-functioning adenoma; PRL: prolactin; TSH: thyroid-stimulating hormone.
MR imaging protocol
MR images were obtained using a 3-T MR scanner (Magnetom Trio, A Tim system T-Class VB17, Siemens, Germany) with a 32-channel phased-array head coil. DWI were obtained in the coronal plane using single-shot echo planar sequences with parallel imaging technique. The imaging parameters of DWI were as follows: repetition time (TR) 2500 ms, echo time (TE) 66 ms, field of view (FOV) 168 × 200 mm, slice thickness 2 mm, intersection gap 0.2 mm, matrix size 128 × 128, 15 averages, b values 0 and 1000 s/mm2, scan time 2 m 47 s. Parallel imaging was performed using generalized autocalibrating parallel acquisitions (GRAPPA), with an acceleration factor of four. ADC maps were automatically calculated. All patients underwent pre-enhanced T1-weighted fluid attenuated inversion recovery imaging (T1-FLAIR) and T2-weighted imaging (T2-WI) and post-enhanced T1-FLAIR in the sagittal and coronal planes after administration of gadolinium 0.1 mmol/kg body weight.
The parameters of T1-FLAIR were TR 2000 ms, TE 11 ms, inversion time 860 ms, flip angle (FA) 130°, echo train length (ETL) 6, and scan time 2 m 54 s. The parameters of T2-WI were TR 4000 ms, TE 76 ms, FA 150°, ETL 12, and scan time 2 m 58 s. All imaging sequences used FOV 200 × 200 mm, slice thickness 2 mm, intersection gap 0.2 mm, and matrix size 320 × 320.
Image analysis
The ADC maps were reviewed using Ziostation 2 (Ziosoft Inc., Tokyo), and placed a volume of interest (VOI) on the whole tumor lesion while superimposing contrast-enhanced MR images by a radiologist (MD, with 12 years of experience in brain MR imaging).
All ADC values within the VOI were generated automatically to compute the average ADC within the tumor. The ADC values were binned to construct the ADC histogram. An example of the process of extracting the VOI of a tumor is shown in Figure 1. The following parameters of the ADC histogram were computed: mean, percentiles (5th, 10th, 25th, 50th, 75th, 90th, and 95th), skewness, and kurtosis. Histogram parameters were then compared between MIB-1 index, Knosp classification as the aggressiveness of pituitary adenomas. Recurrence is another factor of tumor aggressiveness.15 Therefore, histogram parameters were also compared between positive and negative recurrences.
Figure 1.
A 42-year-old woman with prolactin (PRL)-secreting pituitary adenoma. MiB-1 index was 7.6 % and Knosp classification was three. (a) The outline of the tumor is drawn on the contrast-enhanced T1-weighted coronal image. (b) The contours of the tumor of each slice are automatically copied to the exact same location of the corresponding apparent diffusion coefficient (ADC) maps. (c) The data acquired from each slice are summed to generate volume of interest (VOI). (d) The ADC values within the VOI are binned to construct the ADC histogram.
Statistical analysis
Data are expressed as mean ± standard deviation (SD). The Jonkheere-Terpstra test was used to correlate the histogram parameters with the MB-1 index.
The Jonkheere-Terpstra test is a non-parametric test of rank-based trend that is used to determine the significance of trend in more than two independent samples.16 MIB-1 index results were divided into less than one (n = 5), between 1–3 (n = 5), and more than 3 (n = 3). The Mann-Whitney U test was used to compare the histogram parameters between negative and positive invasion. Results of the Knosp classification were divided into two groups of negative (grades 0–2, n = 8) and positive invasion (grades 3–4, n = 5). We also used the Mann-Whitney U test to compare the histogram parameters between positive (n = 3) and negative recurrence (n = 10). A p-value less than 0.05 was considered to be statistically significant. We performed a receiver-operating characteristic (ROC) curve analysis to assess the diagnostic performance of histogram parameters for tumor’s aggressiveness of pituitary adenomas.
All statistical calculations were done with PASW statistical software (ver.23.0, SPSS, IBM, Chicago, Illinois, USA).
Results
The skewness of the ADC histogram showed the only significant difference among the three subgroups of the MIB-1 index (p = 0.030). No significant differences were found in the other ADC histogram parameters (Table 2). All ADC histogram parameters were not significantly different between negative and positive invasion using the Knosp classification (Table 3). Within follow-up periods of 9–57 months, tumor recurrence occurred in three cases. The skewness and kurtosis of the ADC histogram showed significant differences between positive and negative recurrence (skewness p = 0.011, kurtosis p = 0.011) (Table 4). All recurrent tumors showed MIB-1 indices of 3% or more.
Table 2.
Apparent diffusion coefficient (ADC) histogram parameters of MIB-1 index subgroups of pituitary macroadenomas.
| ADC parameter | MIB-1 < 0–1% (n = 5) | 1% MIB-1 < 3% (n = 5) | 3% M1B-1 (n = 3) | p-Value |
|---|---|---|---|---|
| Mean | 815 ± 91 (708–919) | 840 ± 118 (716–1017) | 753 ± 15 (736–767) | 0.470 |
| 5th Percentile | 594 ± 101 (478–695) | 564 ± 105 (478–747) | 563 ± 20 (544–584) | 0.895 |
| 10th Percentile | 642 ± 105 (524–768) | 631 ± 96 (574–803) | 591 ± 22 (573–616) | 0.743 |
| 25th Percentile | 717 ± 76 (627–796) | 720 ± 101 (643–897) | 653 ± 29 (622–680) | 0.264 |
| 50th Percentile | 801 ± 72 (710–875) | 820 ± 114 (712–1016) | 729 ± 35 (689–756) | 0.168 |
| 75th Percentile | 904 ± 84 (797–1000) | 954 ± 151 (786–1129) | 815 ± 25 (786–834) | 0.238 |
| 90th Percentile | 1026 ± 155 (892–1273) | 1088 ± 238 (862–1400) | 931 ± 22 (914–957) | 0.646 |
| 95th Percentile | 1079 ± 171 (931–1361) | 1166 ± 282 (904–1572) | 1072 ± 120 (904–1572) | 0.844 |
| Skewness | 0.097 ± 0.600 (−0.572–1.023) | 0.240 ± 0.406 (−0.321–0.822) | 0.457 ± 0.729 (0.898–2.005) | 0.030a |
| Kurtosis | 0.769 ± 1.168 (−0.49–2.64) | 0.733 ± 0.926 (−0.17–1.86) | 1.380 ± 1.574 (2.04–4.72) | 0.076 |
SD: standard deviation.
Values presented as mean ± SD (range)×10−6 mm2/s.
p < 0.05 (Jonkheere-Terpstra test).
Table 3.
Apparent diffusion coefficient (ADC) histogram parameters of invasion (Knosp classification) of pituitary macroadenomas.
| ADC parameter | Negative, Knosp 0–2 (n = 8) | Positive, Knosp 3–4 (n = 5) | p-Value |
|---|---|---|---|
| Mean | 824 ± 106 (708–1017) | 788 ± 66 (736–894) | 0.622 |
| 5th Percentile | 570 ± 96 (478–747) | 584 ± 76 (497–695) | 0.622 |
| 10th Percentile | 625 ± 88 (524–803) | 627 ± 90 (544–768) | 0.833 |
| 25th Percentile | 711 ± 87 (627–897) | 692 ± 72 (622–796) | 0.724 |
| 50th Percentile | 805 ± 100 (710–1016) | 771 ± 68 (689–866) | 0.622 |
| 75th Percentile | 925 ± 134 (786–1129) | 887 ± 73 (786–976) | 0.524 |
| 90th Percentile | 1066 ± 214 (862–1430) | 967 ± 67 (914–1080) | 0.724 |
| 95th Percentile | 1139 ± 249 (904–1572) | 1066 ± 95 (982–1211) | 0.943 |
| Skewness | 0.366 ± 0.502 (−0.321–1.023) | 0.603 ± 1.053 (−0.572–2.005) | 0.833 |
| Kurtosis | 0.900 ± 0.833 (−0.172–2.041) | 2.148 ± 2.237 (−0.491–4.728) | 0.435 |
SD: standard deviation.
Values presented as mean ± SD (range)×10−6 mm2/s.
Table 4.
Apparent diffusion coefficient (ADC) histogram parameters of recurrence of pituitary macroadenomas.
| ADC parameter | Positive (n = 3) | Negative (n = 10) | p-Value |
|---|---|---|---|
| Mean | 828 ± 100 (708–1017) | 753 ± 15 (736–767) | 0.287 |
| 5th Percentile | 580 ± 98 (478–747) | 563 ± 20 (544–584) | 0.811 |
| 10th Percentile | 636 ± 95 (524–803) | 591 ± 22 (573–616) | 0.692 |
| 25th Percentile | 718 ± 84 (627–897) | 653 ± 29 (622–680) | 0.217 |
| 50th Percentile | 810 ± 90 (710–1016) | 729 ± 35 (689–756) | 0.112 |
| 75th Percentile | 929 ± 118 (786–1129) | 815 ± 25 (786–834) | 0.077 |
| 90th Percentile | 1057 ± 192 (862–1430) | 931 ± 22 (914–957) | 0.489 |
| 95th Percentile | 1122 ± 224 (904–1572) | 1072 ± 120 (991–1211) | 0.937 |
| Skewness | 0.1690 ± 0.4889 (−0.572–1.023) | 1.419 ± 0.556 (0.898–2.005) | 0.014a |
| Kurtosis | 0.751 ± 0.994 (−0.491–2.649) | 3.479 ± 1.353 (2.04–4.72) | 0.014a |
SD: standard deviation.
Values presented as mean ± SD (range) × 10−6 mm2/s.
p< 0.05 (Mann-Whitney U test).
The ROC analysis between positive and negative recurrence showed that both skewness and kurtosis of the ADC achieved area under the curve (AUC) at 0.967, with a cut-off value of 0.86, 100% sensitivity, and 90.0% specificity of skewness 1.954, 100% sensitivity, and 90.0% specificity of kurtosis, respectively (Figure 2).
Figure 2.
Receiver-operating characteristic (ROC) curves of skewness and kurtosis of apparent diffusion coefficient (ADC) in predicting recurrence. The area under the curve (AUC) values were the same for both skewness and kurtosis of ADC (AUC = 0.967, cut-off value of 0.86, 100% sensitivity, and 90.0% specificity of skewness, cut-off value of 1.954, 100% sensitivity, and 90.0% specificity of kurtosis).
Discussion
In this study, the correlation of ADC histogram parameters with MIB-1 index, tumor invasion using Knosp classification, and recurrence were evaluated. Our results showed that skewness and kurtosis had a correlation with MIB-1 index and recurrence. Skewness and kurtosis describe the degree of asymmetry and sharpness of a histogram, respectively. We hypothesize that, rather than showing differences in cell density, aggressive adenomas show a diffusion pattern of water molecules that is more heterogenous and more non-Gaussian.
Previous reports examined the applicability of DWI in the evaluation of the consistency and tumor atypia of pituitary adenomas. Although they suggested that ADC values were useful for differentiation of the consistency and tumor atypia of pituitary adenomas, their conclusions were not different from our results.17–19
In another report using periodically rotated overlapping parallel lines with enhanced reconstruction (PROPELLER) DWI reported that ADC parameters including ADCmin, ADCmean, ADCmax were not correlated with the MIB-1 index.20 To date, histogram analysis of ADC has not been done to evaluate the aggressiveness of pituitary adenomas, therefore this study is the first report to evaluate the aggressiveness of pituitary adenomas using ADC histogram analysis. The important factors in the assessment of the aggressiveness of pituitary adenomas are considered to be tumor subtyping, tumor proliferative potential (mitotic account, MIB-1 index), tumor invasion, and recurrence.
Our study showed that skewness and kurtosis derived from ADC histogram analysis proved to be useful for the evaluation of the aggressiveness of pituitary adenomas.
Our study has several limitations. First, retrospective studies have intrinsic limitations. Second, the study population was small, and it is necessary to increase the number of cases in future studies. Third, selection bias was present. Our study excluded microadenomas with a tumor size less than 10 mm in order to avoid the partial volume effect.
In conclusion, skewness and kurtosis of the ADC histogram were predictive parameters for assessing tumor proliferative potential and recurrence of pituitary adenomas. ADC histogram analysis on the basis of the entire tumor volume is useful in evaluating the aggressiveness of pituitary adenomas.
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.
References
- 1.Priola SM, Esposito F, Cannavo S, et al. Aggressive pituitary adenomas: The dark side of the moon. World Neurosurg 2017; 97: 144–150. [DOI] [PubMed] [Google Scholar]
- 2.Rutkowski MJ, Alward RM, Chen R, et al. Atypical pituitary adenoma: A clinicopathologic case series. J Neurosurg 2018; 128: 1058–1065. [DOI] [PubMed] [Google Scholar]
- 3.Dai C, Feng M, Liu X, et al. Refractory pituitary adenoma: A novel classification for pituitary tumors. Oncotarget 2016; 7: 83657–83668. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Mete O, Lopes MB. Overview of the 2017 WHO classification of pituitary tumors. Endocr Pathol 2017; 28: 228–243. [DOI] [PubMed] [Google Scholar]
- 5.Alimohanmadi M, Ownagh V, Mahouzi L, et al. The impact of immunohistochemical markers of Ki-67 and p53 on the long-term outcome of growth hormone-secreting pituitary adenomas: A cohort study. Asian J Neurosurg 2014; 9: 130–136. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Donati OF, Mazaheri Y, Afaq, et al. Prostate cancer aggressiveness: Assessment with whole-lesion histogram analysis of the apparent diffusion coefficient. Radiology 2014; 271: 143–152. [DOI] [PubMed] [Google Scholar]
- 7.Kyriazi S, Collins DJ, Pennert K, et al. Metastatic ovarian and primary peritoneal cancer: Assessing chemotherapy response with diffusion-weighted MR imaging-Value of histogram analysis of apparent diffusion coefficients. Radiology 2011; 261: 182–192. [DOI] [PubMed] [Google Scholar]
- 8.Mohamed FF, Abouhashem S. Diagnostic value of apparent diffusion coefficient (ADC) in assessment of pituitary macroadenoma consistency. Egypt J Radiol Nucl Med 2013; 44: 617–624. [Google Scholar]
- 9.Kang Y, Choi SH, Kim YJ, et al. Gliomas: Histogram analysis of apparent diffusion coefficient maps with standard-or high-b-value diffusion-weighted MR imaging correlation with tumor grade. Radiology 2011; 261: 882–890. [DOI] [PubMed] [Google Scholar]
- 10.Lee J, Choi SH, Kim Ji-H, et al. Glioma grading using apparent diffusion coefficient map: Application of histogram analysis based on automatic segmentation. NMR Biomed 2014; 27: 1046–1052. [DOI] [PubMed] [Google Scholar]
- 11.Woo S, Cho JY, Kim SY, et al. Histogram analysis of apparent diffusion coefficient map of diffusion-weighted MRI in endometrial cancer: A preliminary correlation study with histological grade. Acta Radiol 2014; 55: 1270–1277. [DOI] [PubMed] [Google Scholar]
- 12.Rosenkrantz AB, Obele C, Rusinek H, et al. Whole-lesion diffusion metrics for assessment of bladder cancer aggressiveness. Abdom Imaging 2015; 40: 327–332. [DOI] [PubMed] [Google Scholar]
- 13.Oner AY, Celik H, Tali T, et al. Thin-section diffusion-weighted magnetic resonance imaging of the brain with parallel imaging. Acta Radiol 2007; 48: 456–463. [DOI] [PubMed] [Google Scholar]
- 14.Knosp E, Steiner E, Kitz K, et al. Pituitary adenomas with invasion of the cavernous sinus space: A magnetic resonance imaging classification compared with surgical findings. Neurosurgery 1993; 33: 610–617. discussion 617–678. [DOI] [PubMed] [Google Scholar]
- 15.Chesney K, Memel Z, Pangal DJ, et al. Variability and lack of prognostic value associated with atypical pituitary adenoma diagnosis: A systematic review and critical assessment of the diagnostic criteria. Neurosurgery 2018; 83: 602–610. [DOI] [PubMed] [Google Scholar]
- 16.Bewick V, Cheek L, Ball J. Statistics review 10: Further nonparametric methods. Critical Care 2004; 8: 196–199. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Zada G, Woodmansee WW, Ramkissoon S, et al. Atypical pituitary adenomas: Incidence, clinical characteristics, and implications. J Neurosurg 2011; 114: 336–344. [DOI] [PubMed] [Google Scholar]
- 18.Tamrazi T, Pekmezci M, Aboian M, et al. Apparent diffusion coefficient and pituitary macroadenomas: Pre-operative assessment of tumor atypia. Pituitary 2016; 20: 195–200. [DOI] [PubMed] [Google Scholar]
- 19.Pierallini A, Caramia F, Falcone C, et al. Pituitary macroadenomas: Preoperative evaluation of consistency with diffusion-weighted MR imaging-initial experience. Radiology 2006; 239: 223–231. [DOI] [PubMed] [Google Scholar]
- 20.Mahmoud OM, Tominaga A, Amatya VJ, et al. Role of propeller diffusion-weighted imaging and apparent diffusion coefficient in the evaluation of pituitary adenomas. Eur J Radiol 2011; 80: 412–417. [DOI] [PubMed] [Google Scholar]