Medial intracranial carotid artery calcifications and hematoma expansion in deep intracerebral hemorrhage Authors: Federico Mazzacane, Stefan Moraru, Beatrice Del Bello, Federica Ferrari, Erica Ferro, Alessandra Persico, Jawed Nawabi, Alessandro Padovani, Anna Cavallini, Andrea Morotti Published: 2024 ABSTRACT AbstractBackgroundMedial intracranial carotid artery calcifications (ICAC) are associated with impaired vascular physiology, increased arterial stiffness and pulse pressure. Their presence might therefore be associated with increased risk of intracerebral hemorrhage (ICH) expansion, according to the avalanche model. We explored the association between ICAC presence and pattern and hematoma expansion (HE).MethodsRetrospective analysis of a monocentric, prospectively collected cohort of ICH patients admitted between June 2017 and October 2023. ICAC pattern was determined by Kockelkoren's rating scale on admission CT; medial ICAC were defined with a >6 points cutoff. A follow‐up CT scan was performed within 72 h. HE was analyzed as a dichotomous (≥6 mL and/or ≥33%) and as a categorical (none/mild/moderate/severe) variable, and its predictors were explored with logistic and ordinal regression respectively, accounting for baseline volume, onset‐to‐CT time, and anticoagulation. All the analyses were stratified by ICH location (supratentorial deep vs lobar ICH).ResultsA total of 201 patients were included (median age 78, 42% females, 59% deep ICH). Medial ICAC were significantly more common in deep ICH with HE compared with non‐expanders (72% vs 49%, p = 0.03), whereas there was no association between ICAC and HE in lobar ICH (53% vs 52%, p = 0.85). This association between medial ICAC and HE in deep ICH remained significant in logistic (aOR 3.11, 95% CI [1.19–9.06], p = 0.03) and ordinal regression (acOR 2.42, 95% CI [1.19–4.99], p = 0.01).InterpretationIpsilateral medial ICAC are associated with higher odds of HE in deep ICH. Our findings are best interpreted as hypothesis generating, requiring prospective validation and further research to characterize the underlying biological mechanisms. Introduction Intracranial carotid artery calcifications (ICAC) are a common finding on non‐contrast head CT (NCCT). 1 Two distinct subtypes of ICAC have been recognized, respectively involving the intimal and medial (internal elastic lamina) layers of the arterial wall, with different underlying pathogenetic mechanisms and risk factors. 2 , 3 Intimal calcifications are closely related to atherosclerosis, whereas medial calcifications are thought to originate from an atherosclerosis‐independent process. 2 , 4 In vivo distinction of ICAC subtypes is possible on NCCT using a histology‐validated rating scale. 5 The medial subtype of arterial calcifications is strongly associated with impaired vessel physiology, leading to increased arterial stiffness and pulse pressure. 6 , 7 , 8 These detrimental effects can determine distal vessels damage, increasing their rupture risk. 9 , 10 Furthermore, medial calcification has been associated with small vessel disease (SVD) burden 11 and appear to be a plausible biological mediator linking hypertension and cerebral SVD. 9 ICAC appear therefore of potential interest in intracerebral hemorrhage (ICH) research, being a plausible biomarker of vessel injury and frailty. ICAC presence might identify patients with a high burden of frail, rupture‐prone cerebral small vessels, therefore contributing to hematoma expansion (HE) according to the avalanche model. 12 , 13 We tested the hypothesis that ICAC presence and pattern identify ICH patients at increased risk of HE. Methods Study population We performed a retrospective analysis of a prospectively collected cohort of ICH admitted at the IRCCS Mondino Stroke Unit in Pavia, Italy, between June 2017 and October 2023. Written informed consent was obtained by patients or waived by the Institutional Review Board. All the study procedures were approved by the local authorities (PN0035588/22). The results of this study are reported according to STROBE guidelines. 14 For the present analysis, we included patients with primary spontaneous intracerebral hemorrhage and (1) baseline non‐contrast CT (NCCT) acquired within 24 h from symptoms onset or time last seen well (LSW), (2) availability of follow‐up NCCT within 72 h from symptoms onset or time LSW, and (3) age >18. Patients with any of the following were excluded: (1) traumatic brain injury, macrovascular lesions or other brain disease underlying the acute ICH, (2) missing or poor quality NCCT imaging, and (3) infratentorial ICH. Clinical variables Clinical data were collected by trained investigators through medical charts, and patients and family members' interviews. ICH care was provided following the AHA/ASA guidelines, including blood pressure lowering, anticoagulation reversal, blood glucose, and temperature management. 15 , 16 The following variables were collected: (1) demographical data: age and sex; (2) medical history: hypertension, diabetes, atrial fibrillation, previous ischemic stroke or ICH, and chronic renal disease; (3) antiplatelet and anticoagulant treatment; (4) index event characteristics: admission Glasgow coma scale score, and systolic and diastolic blood pressure; (5) admission blood test: glucose, coagulation, differential blood cells count, and creatinine. Image analysis Admission NCCT scans were evaluated to define the predominant calcification pattern according to Kockelkoren et al. 5 NCCT slices had 0.5 mm thickness and were evaluated using bone window with multiplanar reconstructions. Points were assigned for circularity (absent: 0, dot(s): 1, <90°: 2, 90–270°: 3, 270°–360°: 4 points), thickness (absent: 0, ≥1.5 mm: 1, <1.5 mm: 3 points), and pattern (indistinguishable: 0, irregular/patchy: 1, continuous: 4 points). Each carotid artery was evaluated separately. Ipsilateral carotid artery to ICH was considered for the analysis. A cutoff of >6 points was used to identify arteries with medial calcification predominance, as previously validated. 5 Illustrative examples of medial and intimal ICAC are provided in Figure 1. Examples of medial (thin, continuous; panels A and B) and intimal (thick discontinuous; panels C and D) internal carotid artery calcifications (ICAC) on non‐contrast computed tomography. All the scans were analyzed by two independent raters (FM and BDB), blinded to the primary outcome of interest (evidence of HE on follow‐up NCCT), and discrepancies were resolved by consensus. Calcification severity was evaluated with the Woodcock scale, and scored as absent, mild (thin and discontinuous calcifications), moderate (thin and continuous or thick discontinuous calcifications), or severe (thick continuous calcifications). 17 Baseline and follow‐up hematoma volumes were obtained by semi‐automatic segmentation using ITK‐SNAP software 18 . The presence of intraventricular hemorrhage was also recorded. HE was defined as volume increase ≥6 mL and/or ≥ 33% from baseline to follow‐up NCCT. 12 , 19 In a secondary analysis, HE was analyzed as a spectrum with different severity degrees as none, mild, moderate, and severe according to validated combined cutoffs of absolute (aHE) and relative HE (rHE): no HE (≤0 mL aHE and ≤0% rHE), mild HE (0 mL < aHE < 6 mL and 0 < rHE < 33%), moderate HE (6 mL ≤ aHE < 12.5 mL and 33% ≤ rHE < 66%), severe HE (aHE ≥12.5 mL or rHE ≥66%). 20 , 21 Statistical analysis Categorical variables were expressed as count (percentage) and compared with χ 2 test. Continuous variables were summarized as mean (SD) or median (interquartile range [IQR]) and compared with t‐test and Mann–Whitney test in case of normal and non‐normal distribution, respectively. Predictors of HE occurrence and HE severity were, respectively, explored with multivariable logistic and ordinal regression models respectively, accounting for baseline ICH volume, time from onset / LSW to baseline NCCT and anticoagulant treatment. 22 All the analyses were stratified by ICH location (supratentorial deep vs lobar) and performed with the statistical package R version 4.4.0 (www.R‐project.org) 23 . p‐values <0.05 were considered statistically significant. Results A total of 201 patients were included (median age 78 years, 42.8% females), of whom 58.7% had a deep ICH. The study cohort characteristics are summarized in Table 1, and Figure 2 illustrates the study selection flowchart. HE occurred in 61 patients (30.3%); 29/118 (24.6%) of deep ICH and 32/83 (38.6%) of lobar ICH. Table 1Patients' characteristics.VariableAll N = 201Deep ICH N = 118Lobar ICH N = 83AllHE N = 61No HE N = 140 p AllHE N = 29No HE N = 89 p AllHE N = 32No HE N = 51 p Age, median (IQR)78 (16)79 (15)78 (17.25)0.0978 (16)79 (15)77 (17)0.3179 (16)79.5 (13.75)78 (17)0.26Sex, female, N (%)86 (42)29 (48)57 (41)0.3747 (40)13 (45)34 (38)0.5339 (47)16 (50)23 (45)0.66Diabetes, N (%)45 (22)18 (30)27 (19)0.1126 (22)8 (28)18 (20)0.4119 (23)10 (31)9 (18)0.15Atrial fibrillation, N (%)45 (22)21 (34)24 (17)0.0125 (21)9 (31)16 (18)0.1420 (24)12 (38)8 (16)0.02Antiplatelet therapy, N (%)66 (33)23 (38)43 (31)0.3331 (26)8 (28)23 (26)0.8535 (42)15 (47)20 (39)0.49Anticoagulant therapy, N (%)41 (20)19 (31)22 (16)0.0125 (21)10 (34)15 (17)0.04416 (19)9 (28)7 (14)0.11Statin therapy, N (%)43 (21)10 (16)33 (24)0.2529 (25)5 (17)24 (27)0.2914 (17)5 (16)9 (18)0.81Admission GCS, median (IQR)15 (2)14 (4)15 (2)0.0215 (1)15 (2)15 (1)0.3614 (4)13 (4)14 (3)0.12Admission systolic blood pressure, median (IQR)165 (35)160 (35)168 (35.5)0.18170 (40)170 (35)170 (40)0.83160 (35)159 (26)160 (34)0.21Onset to NCCT (min), median (IQR)297 (543)223 (286)339 (587.5)0.04240 (506.5)212 (168)305 (540)0.25333 (587)245 (336)435 (680)0.019Baseline hematoma volume (mL), median (IQR)12.28 (26.83)23.85 (47)10.14 (17.06)0.0036.948 (12.201)7.79 (20.23)6.94 (9.97)0.6232.32 (49.27)48.17 (49.74)21.75 (45.59)0.02Baseline IVH presence, N (%)69 (34)18 (30)51 (36)0.3445 (38)9 (31)36 (40)0.3724 (29)9 (28)15 (29)0.9Follow‐up hematoma volume (mL), median (IQR)12.46 (33.24)38.37 (77.07)9.87 (15.67)<0.0017.523 (12.43)12.12 (32.18)6.88 (10.28)0.00635.58 (55.70)63.21 (63.38)22.14 (39.4)<0.001Follow‐up IVH presence, N (%)77 (38)28 (46)49 (35)0.1445 (38)13 (45)32 (36)0.3932 (39)15 (47)17 (33)0.22INR, median (IQR)1.01 (0.13)1.04 (0.23)1 (0.1)0.0061 (0.15)1.08 (0.32)0.99 (0.09)0.0121.02 (0.12)1.04 (0.15)1 (0.1)0.2Admission glucose, median (IQR)120 (43)120 (44.5)122 (42)0.92117 (42)114.5 (31.75)124 (41.5)0.12124.5 (43)138 (54.25)120 (37.25)0.16Platelets, median (IQR)210 (76)194 (55)214.5 (80.5)0.14209 (79.25)197 (64)212 (83)0.4210 (74)194 (48.75)216 (87)0.18Ipsilateral medial ICAC, N (%)108 (54)38 (62)70 (50)0.1165 (55)21 (72)44 (49)0.03143 (52)17 (53)26 (51)0.85Moderate/severe ICAC, N (%)117 (59)41 (67)76 (54)0.0968 (58)20 (69)48 (54)0.1649 (59)21 (66)28 (55)0.33ICAC, intracranial carotid artery calcifications; INR, international normalized ratio; IVH, intraventricular hemorrhage; NCCT, non‐contrast computed tomography. Patients' characteristics. ICAC, intracranial carotid artery calcifications; INR, international normalized ratio; IVH, intraventricular hemorrhage; NCCT, non‐contrast computed tomography. Patients' selection flowchart. ICH, intracerebral hemorrhage; NCCT, non‐contrast computed tomography. Ipsilateral medial ICAC were present in 108 (53.7%) patients, and ICAC severity according to Woodcock classification was moderate–severe in 117 (58.2%) patients. The prevalence of ipsilateral medial ICAC (55.1% vs 51.8%, p = 0.65) and moderate–severe ICAC (57.6 vs 59.0%, p = 0.84) was similar between patients with deep and lobar ICH. Patients with deep ICH experiencing HE had more frequent ipsilateral medial ICAC compared with non‐expanders (72.4% vs 49.4%, p = 0.03), while the prevalence of medial ICAC according to HE occurrence was similar in subjects with lobar ICH (53.1% vs 51.0%, p = 0.85). Moderate–severe ICAC frequency was similar in expanders and non‐expanders in both locations (deep ICH, 68.9 vs 53.9%, p = 0.16 and lobar ICH 65.6 vs 54.9%, p = 0.33). A positive association between the presence of medial ICAC and deep HE was found in univariate logistic regression (OR 2.68, 95% CI [1.11–7.04], p = 0.03) and remained significant when accounting for confounders (adjusted OR 3.11, 95% CI [1.19–9.06], p = 0.03), as summarized in Table 2 and Figure 3A. Table 2Logistic regression models result. Outcome: hematoma expansion ≥33% or 6 mL.UnivariateMultivariate a ORLower 95% CIUpper 95% CI p‐valueaORLower 95% CIUpper 95% CI p‐valueDeep ICHIpsilateral medial ICAC2.681.117.040.033.111.199.060.03Baseline ICH volume (10 mL increase)1.321.001.790.061.41.032.000.04Onset‐to‐CT time (h)0.960.881.020.250.950.871.030.23Anticoagulant therapy2.60.996.680.052.010.715.510.18Lobar ICHIpsilateral medial ICAC1.090.452.660.851.070.422.770.89Baseline ICH volume (10 mL increase)1.120.991.280.081.120.981.280.1Onset‐to‐CT time (h)0.920.840.990.030.920.840.990.04Anticoagulant therapy2.460.827.720.112.170.687.150.19ICAC, intracranial carotid artery calcifications; ICH, intracerebral hemorrhage. a Included variables: ipsilateral medial ICAC, baseline hematoma volume, onset‐to‐CT time, and anticoagulant therapy. Logistic regression models result. Outcome: hematoma expansion ≥33% or 6 mL. ICAC, intracranial carotid artery calcifications; ICH, intracerebral hemorrhage. Included variables: ipsilateral medial ICAC, baseline hematoma volume, onset‐to‐CT time, and anticoagulant therapy. Hematoma expansion (HE) expansion and severity in deep and lobar intracerebral hemorrhage (ICH) according to presence of ipsilateral medial intracranial carotid artery calcifications (ICAC). Ordinal regression showed an association between ipsilateral medial ICAC and a shift toward more severe HE in deep ICH in univariate (cOR 2.24, 95% CI [1.14–4.48], p = 0.02) and multivariable (acOR 2.42, 95% CI [1.19–4.99], p = 0.01) models (Table 3 and Fig. 3B). Table 3Ordinal regression models result. Outcome: HE severity category, combined rHE and aHE cutoffs (0/6/12.5 mL – 0%/33%/66%).UnivariateMultivariate a cORLower 95% CIUpper 95% CI p‐valueacORLower 95% CIUpper 95% CI p‐valueDeep ICHIpsilateral medial ICAC2.241.144.480.022.421.194.990.01Baseline ICH volume (10 mL increase)1.311.011.720.051.371.041.830.02Onset‐to‐CT time (h)0.970.921.020.280.970.921.020.22Anticoagulant therapy2.070.924.730.081.630.73.780.25Lobar ICHIpsilateral medial ICAC1.10.52.420.811.030.462.310.94Baseline ICH volume (10 mL increase)1.120.991.270.061.10.981.250.11Onset‐to‐CT time (h)0.940.881.000.040.950.891.010.09Anticoagulant therapy2.170.825.840.121.880.75.120.21ICAC, intracranial carotid artery calcifications; ICH, intracerebral hemorrhage. a Included variables: ipsilateral medial ICAC, baseline hematoma volume, onset‐to‐CT time, and anticoagulant therapy. Ordinal regression models result. Outcome: HE severity category, combined rHE and aHE cutoffs (0/6/12.5 mL – 0%/33%/66%). ICAC, intracranial carotid artery calcifications; ICH, intracerebral hemorrhage. Included variables: ipsilateral medial ICAC, baseline hematoma volume, onset‐to‐CT time, and anticoagulant therapy. No association between the presence of ipsilateral medial ICAC and HE (adjusted OR 1.07, 95% CI [0.42–2.77], p = 0.89) or HE severity shift (acOR 1.03, 95% CI [0.46–2.31], p = 0.94) was observed in patients with lobar ICH (Tables 2 and 3; Fig. 3). Discussion The main finding of our analysis is the association between ipsilateral medial ICAC presence, the occurrence of HE, and its severity. We have also reported an interaction between medial ICAC presence, HE and ICH location, with ICAC being associated with higher odds of HE only in patients with deep ICH. Previous literature on the significance of ICAC patterns in ICH is scarce, and only a previous study on the topic suggested a potential association between medial ICAC and HE, although this finding was not confirmed after adjusting for potential counfounders. 24 This non‐significant result may have been driven by the lack of ICH location stratification. Our findings are best interpreted as hypothesis generating and might provide novel insights into the pathophysiology of HE. The location‐specific impact of ICAC on HE risk might reflect the different risk factors and underlying small vessel disease of deep and lobar ICH, namely arteriolosclerosis and cerebral amyloid angiopathy. 25 , 26 Medial ICAC appear to be a reliable biomarker of small vessel frailty and the biological mediator of the association between hypertension 9 , the main risk factor for deep ICH, and cerebral small vessel disease. 25 ICAC presence might therefore indicate the presence of a high burden of stiff and rupture‐prone small vessels, contributing to HE according to the avalanche model, where the mass effect of the initial hematoma leads to multiple surrounding small vessels' rupture and further bleeding (Fig. 4). 12 Conversely, although cortical and juxtacortical small vessels could be affected by hypertensive small vessel disease, it is plausible that the extent of the damage in these regions is less severe, and therefore insufficient to promote lobar ICH expansion. 27 , 28 , 29 , 30 Another potential mediator of the association between medial ICAC and HE in deep ICH is increased pulse pressure, arising from increased arterial stiffness. In line with this hypothesis, a recent study demonstrated an association between higher pulse pressure and increased odds of HE. 31 While these mechanisms could potentially explain the association, the lack of CT angiography of the epiaortic and intracranial vessels limits our ability to investigate whether flow‐disturbing carotid or intracranial stenosis in patients with ICAC contributes to this relationship. Additionally, blood flow characteristics (e.g., via transcranial Doppler) were not assessed in our study, preventing us from examining the impact of medial ICAC on cerebral blood flow and its potential role in the association between ICAC and HE. These research questions warrant further investigation. Potential mechanisms involved in the association between medial intracranial carotid arteries calcification (ICAC) and deep ICH hematoma expansion (created with BioRender.com). Medial ICAC are associated with a higher small vessel disease burden and arterial stiffness, the latter leading to increased arterial pulse pressure. The combination of damaged rupture‐prone small vessels and detrimental hemodynamic changes may facilitate hematoma expansion. Our results might have relevant implications for clinical practice and future studies, as accurate prediction of HE remains a clinical research priority and the predictive performance of currently available tools remains suboptimal 12 , 32 . ICAC can be rated on NCCT, a low cost widely available tool and their assessment might improve stratification of HE risk in clinical practice and future trials. Some limitations should be considered in the interpretation of our findings. First, selection bias might have influenced our analyses. Second, the monocentric nonrandomized and retrospective design of our study might have led to confounding by unmeasured variables. Third, we were unable to precisely account for blood pressure variability, magnitude of SBP reduction and timing of coagulopathy reversal in the acute phase. 33 Finally, we were unable to measure and account for SVD burden because of the absence of MRI data. Conclusions Ipsilateral medial ICAC were associated with HE occurrence and a shift toward more severe HE in patients with deep ICH. Further studies appear warranted to validate our results and characterize the underlying biological mechanisms. Conflict of Interest Nothing to report. Author Contributions F.M., A.M., A.C., J.N., and A.Pa. contributed to the conception and design of the study. F.M., A.M., S.M., B.D.B., F.F., A.Pe., and E.F. contributed to the acquisition and analysis of data. F.M., A.M., S.M., and B.D.B. contributed to drafting the text or preparing the figures. Acknowledgments This work was supported by the Italian Ministry of Health “Ricerca Corrente 2022‐2024” granted to IRCCS Mondino Foundation. Funding Statement This work was funded by Ministero della Salute grant Ricerca Corrente 2022‐2024. Data Availability Statement The data that support the findings of this study are available in Zenodo, at http://doi.org/10.5281/zenodo.13134015. References Associated Data Data Availability Statement The data that support the findings of this study are available in Zenodo, at http://doi.org/10.5281/zenodo.13134015.