Diffuse Patterns of Nonaneurysmal Subarachnoid Hemorrhage Originating from the Basal Cisterns Have Predictable Vasospasm Rates Similar to Aneurysmal Subarachnoid Hemorrhage Brian P. Walcott, MD,* Christopher J. Stapleton, MD,* Matthew J. Koch, MD,* and Christopher S. Ogilvy, MD†‡

Background: Nonaneurysmal subarachnoid hemorrhage (SAH) has been historically associated with a benign clinical course. However, recent studies have suggested that nonaneurysmal SAH can present with different hemorrhage patterns that may be associated with differential rates of morbidity. Herein, we analyze a retrospective consecutive cohort of patients with nonaneurysmal SAH to determine outcomes. We also seek to evaluate a validated radiographic grading scale to determine its utility in predicting vasospasm in the setting of different hemorrhage patterns. Methods: After institutional review board approval, the records of 563 consecutive patients admitted with spontaneous SAH between January 2007 and 2014 were retrospectively reviewed. A total of 138 of these patients had no identifiable source of hemorrhage and were further divided into 2 groups depending on their pattern of hemorrhage: perimesencephalic or diffuse. Clinical characteristics and outcomes were assessed. Results: In nonaneurysmal SAH, 70 patients (50.7%) had a perimesencephalic pattern of hemorrhage, whereas 68 (49.3%) experienced diffuse SAH. Radiographic vasospasm developed in 6 patients (8.6%) with perimesencephalic SAH and in 14 patients (20.6%) with a diffuse SAH pattern. When comparing historical rates of vasospasm based on the Barrow Neurological Institute (BNI) scale and rates in the nonaneurysmal diffuse pattern in this series, there was no significant difference in distribution (chi-square; P 5 .149), compared with a difference seen with the perimesencephalic group (P , .00001). Conclusions: Nonaneurysmal SAH is associated with the potential for vasospasm, with higher rates in the diffuse versus perimesencephalic SAH patterns. The BNI grading scale for aneurysmal SAH can be used to predict the risk of vasospasm in diffuse, nonaneurysmal SAH. Key Words: Aneurysm—vasospasm—subarachnoid hemorrhage—hydrocephalus. Ó 2015 by National Stroke Association

Spontaneous subarachnoid hemorrhage (SAH) can be classified based on etiology: aneurysmal or nonaneurysmal. Aneurysmal SAH has well-recognized outcomes,

and treatment guidelines are focused on early aneurysm obliteration to prevent re-rupture, as well as the prevention of secondary insults that result from the sequelae of

From the *Department of Neurosurgery, Massachusetts General Hospital and Harvard Medical School, Boston, MA; †Division of Neurosurgery, Department of Surgery, Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, MA; and ‡Brain Aneurysm Institute, Beth Israel Deaconess Medical Center, Boston, MA. Received September 23, 2014; revision received October 22, 2014; accepted November 14, 2014. B.P.W., C.J.S., M.J.K., and C.S.O. declare that they have no conflicts of interest.

Address correspondence to Brian P. Walcott, MD, Department of Neurosurgery, Massachusetts General Hospital, 15 Parkman Street, ACC-021, Boston, MA 02114. E-mail: walcott.brian@mgh. harvard.edu. 1052-3057/$ - see front matter Ó 2015 by National Stroke Association http://dx.doi.org/10.1016/j.jstrokecerebrovasdis.2014.11.011

Journal of Stroke and Cerebrovascular Diseases, Vol. 24, No. 4 (April), 2015: pp 795-801

795

B.P. WALCOTT ET AL.

796 1

the initial SAH such as vasospasm and hydrocephalus. The outcomes of nonaneurysmal SAH are less well known, and as a result, their management is less standardized. Repeated neurovascular imaging is often performed to identify an aneurysmal source that may not have been apparent on initial studies.2,3 When spontaneous SAH is seen without an aneurysmal source, hemorrhage patterns can be dichotomized into either the perimesencephalic pattern (Figure 1),4 where blood is confined mainly to the basal cisterns surrounding the midbrain, or diffuse (Figure 2), where blood extends into the distal Sylvian or interhemispheric fissures. Although nonaneurysmal hemorrhage has typically been associated with a benign course,5-8 recent evidence has suggested that hydrocephalus, vasospasm, and stroke can all occur.9,10 We sought to identify outcomes from a series of patients with nonaneurysmal SAH and to determine the utility of a modern radiographic grading scale to predict rates of vasospasm.

Methods Patient Characteristics After institutional review board approval, the records of 563 consecutive patients admitted with spontaneous SAH between January 2007 and January 2014 were retro-

spectively reviewed. Patients with SAH limited to the convexity of the cortical surface were excluded. Of the 563 patients, 138 were further selected who did not have evidence of an aneurysmal source. All historical, clinical, radiographic, and follow-up information was obtained retrospectively from the electronic medical record. Patient clinical characteristics were collected and included age, sex, prehemorrhage modified Rankin Scale score,11 Hunt and Hess grade,12 and length of follow-up. Radiographic characteristics of the initial hemorrhage were also identified, including the modified Fisher grade,13 the Barrow Neurological Institute (BNI) grade,14 presence of intraventicular hemorrhage, and the need for temporary external ventriculostomy. Radiographic grading scale scores were determined by the authors, who were blinded to the clinical outcome of study patients.

SAH Management All patients were initially managed in a dedicated neurosciences intensive care unit. Patients with clinical or radiographic hydrocephalus or with admission Hunt and Hess grades 3 or more underwent external ventricular drain placement for cerebrospinal fluid (CSF) diversion and intracranial pressure monitoring. External ventricular drains were managed according to a

Figure 1. Perimesencephalic subarachnoid hemorrhage pattern. Noncontrast axial head computed tomography of patient who experienced a nonaneurysmal subarachnoid hemorrhage confined to mainly the basal cisterns surrounding the midbrain. Representative images are demonstrated in sequence from A-D in their inferior to superior location.

VASOSPASM IN NONANEURYSMAL SAH

797

Figure 2. Diffuse subarachnoid hemorrhage pattern. Noncontrast axial head computed tomography of a patient who experienced a nonaneurysmal subarachnoid hemorrhage where blood extended into the distal Sylvian and interhemistpheric fissures. Representative images are demonstrated in sequence from A-D in their inferior to superior location.

previously published protocol.15 All patients received oral nimodipine for vasospasm prevention, although a small percentage of patients (not recorded) could not tolerate it secondary to nimodipine-related hypotension. Depending on provider preference, transcranial Doppler (TCD) ultrasonography was performed for vasospasm surveillance. Computed tomography (CT) angiography was performed on patients with elevated or rising TCD velocities or with a change in neurologic status thought to be attributable to vasospasm. Diagnostic cerebral angiography was performed on patients with vasospasm seen on CT angiography or on patients in whom there was a high clinical suspicion for vasospasm despite a negative CT angiogram. All patients with vasospasm received hypertensive therapy before initiation of endovascular intervention.

Vasospasm Vasospasm was defined as new or progressive intracranial arterial narrowing (mild, moderate, or severe) identified on catheter or CT angiography.14,16 The presence and degree of vasospasm was previously determined by the radiologist in the radiology interpretation report. Elevated TCD blood flow velocity measurements often

prompted vascular imaging, but sonographic data were not used to define vasospasm. Symptomatic vasospasm was defined as neurologic symptoms referable to a region of angiographic vasospasm in the absence of alternative explanations. The determination of symptomatic vasospasm was made during the patient’s hospital stay in a nonblinded manner by the staff neurosurgeon and the neurocritical care physician of record.

Outcomes Outcomes extracted from the medical record included discharge disposition (home, inpatient rehabilitation, or in-hospital mortality), the development and degree of radiographic vasospasm, the development of clinical vasospasm, whether any endovascular vasospasm treatment was performed, whether the patient required permanent CSF diversion (ventriculoperitoneal shunting), and modified Rankin scale score at 1 year posthemorrhage. Other outcomes included length of hospital stay, number of inpatient catheter angiograms, number of inpatient neurovascular imaging studies (magnetic resonance angiography or CT angiography), maximum TCD velocities, number of days TCD velocities exceeded more than 200 cm/second, and maximum Lindegaard ratio.17

B.P. WALCOTT ET AL.

798

Table 1. Patient characteristics Total, N 5 138 (100%)

Characteristics Age, mean (SEM; range) Sex Male Female Hunt–Hess grade 1 2 3 4 5 Prehemorrhage mRS* 0 1 2 3 4 Length of follow-up, mean (SEM; range)

Perimesencephalic hemorrhage, n 5 70 (50.7%)

Diffuse hemorrhage, n 5 68 (49.3%)

54.9 (1.7; 4-89)

56.3 (1.7; 24-86)

74 (53.6) 64 (46.4)

36 34

38 30

26 (18.8) 89 (64.5) 17 (12.3) 3 (2.2) 3 (2.2)

17 47 5 1 0

9 42 12 2 3

67 1 0 1 1 220.3 (45.6; 0-1969)

56 6 3 2 1 241.7 (46.4; 2-1642)

55.6 (1.2; 4-89)

123 (89.1) 7 (5.1) 3 (2.2) 3 (2.2) 2 (1.4) 230.8 (32.4; 0-1969)

Abbreviation: SEM, standard error of the mean. Values represent number of patients (%), unless indicated otherwise. *No patients with modified Rankin Scale (mRS) score 5 or 6.

Statistical Analysis Descriptive statistics were calculated for clinical and radiographic factors. BNI grading scale scores were compared between previously published rates of vasospasm for different grades18 and our patients using the

chi-square test for trend. Our null hypothesis was that there was no significant difference in the distribution of vasospasm rates based on radiographic grading scale between aneurysmal and nonaneurysmal hemorrhage. Analyses were performed using Stata MP (version 13; Stata, College Station, TX) with an a level set at .05.

Table 2. Hemorrhage characteristics

Characteristics

Perimesencephalic hemorrhage, n 5 70 (50.7%)

Diffuse hemorrhage, n 5 68 (49.3%)

5 (3.6) 18 (13.0) 1 (.7) 67 (48.6) 47 (34.1)

5 6 1 41 18

0 12 0 26 29

6 (4.3) 59 (42.8) 56 (40.6) 13 (9.4) 4 (2.9)

5 31 33 1 0

1 28 23 12 4

90 (65.2) 48 (34.8)

51 19

39 29

119 (86.2) 19 (13.8)

67 3

52 16

Total, N 5 138 (100%)

Modified Fisher grade 0 1 2 3 4 BNI grade 1 2 3 4 5 Intraventricular hemorrhage no yes Ventriculostomy no yes

Abbreviation: SEM, standard error of the mean. Values represent number of patients (%), unless indicated otherwise.

VASOSPASM IN NONANEURYSMAL SAH

799

Table 3. Outcomes

Outcomes Discharge disposition Home Inpatient rehabilitation Dead Radiographic vasospasm No Yes Degree of vasospasm None Mild Moderate Severe Clinical vasospasm No Yes Endovascular intervention no Yes VP shunt No Yes mRS at 1 y 0 1 2 3 4 5 6 Length of stay, mean (SEM; range), d Catheter angiograms, mean (SEM; range), number Noninvasive neurovascular imaging, mean (SEM; range), # CTA and MRA Transcranial Doppler, mean (SEM; range), number Maximum transcranial Doppler velocity, mean (SEM; range), cm/second Number of Days transcranial Doppler velocity more than 200 cm/second, mean (SEM; range) Maximum Lindegaard ratio, mean (SEM; range)

Perimesencephalic hemorrhage, n 5 70 (50.7%)

Diffuse hemorrhage, n 5 68 (49.3%)

113 (81.9) 23 (16.7) 2 (1.4)

64 6 0

49 17 2

118 (85.5) 20 (14.5)

64 6

54 14

118 (85.5) 7 (5.1) 5 (3.6) 8 (5.8)

64 2 2 2

55 4 3 6

132 (95.7) 6 (4.3)

70 0

62 6

135 (97.8) 3 (2.2)

70 0

65 3

133 (96.4) 5 (3.6)

70 0

63 5

80 (58.0) 33 (23.9) 11 (8.0) 6 (4.3) 4 (2.9) 1 (.7) 3 (2.2) 9.6 (0.6; 1-37) 1.3 (.05; 1-4) 2.1 (0.1; 1-5)

51 12 2 3 1 0 1 7.1 (.5; 2-30) 1.2 (.06; 1-4) 2.2 (.1; 1-5)

29 21 9 3 3 1 2 12.1 (1.1; 1-37) 1.4 (.07; 1-4) 2.1 (.1; 1-5)

3.8 (.3; 0-21) 146.0 (5.8; 40-367)

2.8 (.3; 0-10) 133.1 (6.1; 40-323)

4.9 (.6; 0-21) 158 .7 (9.5; 60-367)

.7 (.2; 0-16)

.3 (.15; 0-6)

1.14 (.40; 0-16)

3.2 (.5; 1.5-5.7)

5.3 (.9; 1.8-12.2)

Total, N 5 138 (100%)

4.4 (.6; 1.5-12.2)

Abbreviations: CTA, computed tomography angiography; MRA, magnetic resonance angiography; mRS, modified Rankin Scale; SEM, standard error of the mean; VP, ventriculoperitoneal. Values represent number of patients (%), unless indicated otherwise.

Results A total of 138 patients were identified with nonaneurysmal SAH with a mean follow-up of 230.8 days (standard error of the mean, 32.4; range, 0-1969). (Table 1) There were no re-hemorrhage events in the follow-up period for any patient. Perimesencephalic hemorrhage accounted for 70 cases (50.7%), whereas diffuse hemorrhage was seen in 68 patients (49.3%). (Table 2) Application of the modified Fisher scale13 identified a dis-

tribution of hemorrhage patterns, with most patients being grade 3 (n 5 67, 48.6%). Likewise, application of the BNI scale identified a distribution of hemorrhage patterns, with the majority being grade 2 (n 5 59, 42.8%) or grade 3 (n 5 56, 40.6%). Of the 138 patients, 20 patients (14.5%) developed radiographic vasospasm, which was more common in the diffuse hemorrhage group (n 5 14, 20.6%) compared with the perimesencephalic group (n 5 6, 8.6%). (Table 3) There were no patients in the perimesencephalic

B.P. WALCOTT ET AL.

800

group who developed symptomatic (clinical) vasospasm, whereas 6 patients in the diffuse hemorrhage group did. A total of 5 patients required permanent CSF diversion, all of which occurred after diffuse hemorrhage. Two patients died, both of whom were classified as having a diffuse hemorrhage pattern. When comparing historical rates of radiographic vasospasm based on the BNI scale and rates in the nonaneurysmal diffuse pattern in this series, there was no significant difference in distribution rates within grade in the diffuse hemorrhage group (chi-square for trend; P 5.149), whereas the distributions were significantly different with the perimesencephalic group (chi-square for trend; P , .00001).

Discussion After aneurysmal SAH, aneurysm re-rupture, vasospasm, and hydrocephalus are significant causes of morbidity and mortality.14,19-22 Early management is aimed at mitigating these risks, and guidelines have been adopted, which help to standardize the approach to these issues.1 However, less is known about spontaneous nonaneurysmal SAH, making management decisions challenging at times. Many theories regarding the etiology of these hemorrhages exist, including tearing of venous structures or rupture of fine arterial vessels adjacent to the mesencephalon. The results of this study provide some justification for what our own clinical practice pattern has been: vigilant, intensive care monitoring for vasospasm development in patients with diffuse hemorrhage patterns and expedited hospital discharge for patients after negative digital subtraction angiography with perimesencephalic hemorrhage patterns. The rate of shunt-dependent hydrocephalus (7.3%) also underscores the potential nonbenign clinical course in diffuse SAH. Although some SAH radiographic grading scales are nominal, such as the modified Fisher scale, we elected to apply an ordinal scale (BNI scale) to our study population given the hypothesis that the pathophysiology of vasospasm is correlated to the volume of blood products released into the subarachnoid space. We found that the nonaneurysmal diffuse hemorrhage group did follow a trend of similar distributions of vasospasm as SAH patterns seen in patients with ruptured aneurysms. However, when applied to patients with perimesencephalic hemorrhage patterns, the vasospasm rates were significantly different from those observed in aneurysmal hemorrhage. This supports an alternative hypothesis that incorporates the etiology of the hemorrhage as an important factor in the pathophysiologic pathway of vasospasm.23-25 Regardless of the etiology of a spontaneous SAH, we have followed a consistent practice of primary investigation with CT angiography to identify any aneurysmal source.26 When results are negative or equivocal, catheter-based digital subtraction angiography is per-

formed, as was the case in all patients in this series. Based on clinical suspicion, negative results of these 2 studies are sometimes investigated with further neurovascular imaging to increase diagnostic yield.3,27 Patients in this series had all had negative initial neurovascular imaging studies, although the variability of subsequent follow-up studies performed is a potential confounding factor. We feel that all patients represented in this series represent true nonaneurysmal hemorrhage, as was determined during the time of the initial hospitalization according to this protocol, augmented by any delayed outpatient imaging studies. All forms of spontaneous SAH warrant investigation to rule out an aneurysmal source, as well as some level of clinical suspicion for the development of vasospasm. The optimal level of surveillance for vasospasm is not determined by this study, but the relatively high rate of radiographic vasospasm (20.6%) in the diffuse hemorrhage pattern group suggests the utility of intensive clinical monitoring. Furthermore, although the rate of radiographic vasospasm in perimesencephalic hemorrhage group was low, it was not zero. Even when vasospasm does occur, there is debate over a direct correlation of vasospasm with delayed clinical ischemia.28-30 Given the resource utilization employed to treat non-aneurysmal SAH, future prospective study is necessary to establish the efficacy of intensive surveillance and management.

Conclusion Nonaneurysmal SAH is associated with the potential for vasospasm, with higher rates observed in the diffuse versus perimesencephalic SAH pattern. The BNI grading scale for aneurysmal SAH can be used to predict the risk of vasospasm in diffuse, nonaneurysmal SAH.

References 1. Connolly ES, Rabinstein AA, Carhuapoma JR, et al. Guidelines for the management of aneurysmal subarachnoid hemorrhage a guideline for healthcare professionals from the American Heart Association/American Stroke Association. Stroke 2012;43:1711-1737. 2. Dalyai R, Chalouhi N, Theofanis T, et al. Subarachnoid hemorrhage with negative initial catheter angiography: a review of 254 cases evaluating patient clinical outcome and efficacy of short- and long-term repeat angiography. Neurosurgery 2013;72:646-652. 3. Grannan BL, Yanamadala V, Walcott BP, et al. Repeated neurovascular imaging in subarachnoid hemorrhage when initial studies are negative. J Clin Neurosci 2014; 21:993-996. 4. Rinkel G, Wijdicks E, Vermeulen M, et al. Nonaneurysmal perimesencephalic subarachnoid hemorrhage: CT and MR patterns that differ from aneurysmal rupture. Am J Neuroradiol 1991;12:829-834. 5. Van Gijn J, Van Dongen K, Vermeulen M, et al. Perimesencephalic hemorrhage: a nonaneurysmal and benign form of subarachnoid hemorrhage. Neurology 1985; 35:493-497.

VASOSPASM IN NONANEURYSMAL SAH 6. Rinkel GJ, Wijdicks EF, Vermeulen M, et al. The clinical course of perimesencephalic nonaneurysmal subarachnoid hemorrhage. Ann Neurol 1991;29:463-468. 7. Rinkel G, Wijdicks E, Vermeulen M, et al. Outcome in perimesencephalic (nonaneurysmal) subarachnoid hemorrhage: a follow-up study in 37 patients. Neurology 1990;40:1130-1132. 8. Schwartz TH, Solomon RA. Perimesencephalic nonaneurysmal subarachnoid hemorrhage: review of the literature. Neurosurgery 1996;39:433-440. 9. Whiting J, Reavey-Cantwell J, Velat G, et al. Clinical course of nontraumatic, nonaneurysmal subarachnoid hemorrhage: a single-institution experience. Neurosurg Focus 2009;26:E21. 10. Lin N, Zenonos G, Kim AH, et al. Angiogram-negative subarachnoid hemorrhage: relationship between bleeding pattern and clinical outcome. Neurocrit Care 2012; 16:389-398. 11. Van Swieten J, Koudstaal P, Visser M, et al. Interobserver agreement for the assessment of handicap in stroke patients. Stroke 1988;19:604-607. 12. Hunt WE, Hess RM. Surgical risk as related to the time of intervention in the repair of intracranial aneurysms. Neurosurgical Classics II 1967;342-346. 13. Frontera JA, Claassen J, Schmidt JM, et al. Prediction of symptomatic vasospasm after subarachnoid hemorrhage: the modified Fisher scale. Neurosurgery 2006; 59:21-27. discussion 21–27. 14. Hoh BL, Kleinhenz DT, Chi YY, et al. Incidence of ventricular shunt placement for hydrocephalus with clipping versus coiling for ruptured and unruptured cerebral aneurysms in the Nationwide Inpatient Sample database: 2002 to 2007. World Neurosurg 2011;76:548-554. 15. Klopfenstein JD, Kim LJ, Feiz-Erfan I, et al. Comparison of rapid and gradual weaning from external ventricular drainage in patients with aneurysmal subarachnoid hemorrhage: a prospective randomized trial. J Neurosurg 2004;100:225-229. 16. Sheth SA, Hausrath D, Numis AL, et al. Intraoperative rerupture during surgical treatment of aneurysmal subarachnoid hemorrhage is not associated with an increased risk of vasospasm. J Neurosurg 2014;120: 409-414. 17. Lindegaard KF, Nornes H, Bakke S, et al. Cerebral vasospasm after subarachnoid haemorrhage investigated by means of transcranial Doppler ultrasound. Proceedings of the 8th European Congress of Neurosurgery Barcelona, September 6–11, 1987. Vienna: Springer 1988:81-84. 18. Wilson DA, Nakaji P, Abla AA, et al. A simple and quantitative method to predict symptomatic vasospasm after

801

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

subarachnoid hemorrhage based on computed tomography: beyond the Fisher scale. Neurosurgery 2012; 71:869-875. Velat GJ, Kimball MM, Mocco JD, et al. Vasospasm after aneurysmal subarachnoid hemorrhage: review of randomized controlled trials and meta-analyses in the literature. World Neurosurg 2011;76:446-454. Inagawa T, Kamiya K, Ogasawara H, et al. Rebleeding of ruptured intracranial aneurysms in the acute stage. Surg Neurol 1987;28:93-99. Bailes JE, Spetzler RF, Hadley MN, et al. Management morbidity and mortality of poor-grade aneurysm patients. J Neurosurg 1990;72:559-566. Kassell N, Sasaki T, Colohan A, et al. Cerebral vasospasm following aneurysmal subarachnoid hemorrhage. Stroke 1985;16:562-572. Rinkel G, Van Gijn J, Wijdicks E. Subarachnoid hemorrhage without detectable aneurysm. A review of the causes. Stroke 1993;24:1403-1409. Wijdicks EF, Schievink WI. Perimesencephalic nonaneurysmal subarachnoid hemorrhage: first hint of a cause? Neurology 1997;49:634-636. Schievink WI, Wujicks EF. Origin of pretruncal nonaneurysmal subarachnoid hemorrhage: ruptured vein, perforating artery, or intramural hematoma? Mayo Clin Proc 2000;75:1169-1173. Hoh BL, Cheung AC, Rabinov JD, et al. Results of a prospective protocol of computed tomographic angiography in place of catheter angiography as the only diagnostic and pretreatment planning study for cerebral aneurysms by a combined neurovascular team. Neurosurgery 2004; 54:1329-1342. Topcuoglu MA, Ogilvy CS, Carter BS, et al. Subarachnoid hemorrhage without evident cause on initial angiography studies: diagnostic yield of subsequent angiography and other neuroimaging tests. J Neurosurg 2003; 98:1235-1240. Vergouwen MD, Ilodigwe D, Macdonald RL. Cerebral infarction after subarachnoid hemorrhage contributes to poor outcome by vasospasm-dependent and -independent effects. Stroke 2011;42:924-929. Diringer MN. Controversy: does prevention of vasospasm in subarachnoid hemorrhage improve clinical outcome? Stroke 2013;44:S29-S30. Etminan N, Vergouwen MD, Ilodigwe D, et al. Effect of pharmaceutical treatment on vasospasm, delayed cerebral ischemia, and clinical outcome in patients with aneurysmal subarachnoid hemorrhage: a systematic review and meta-analysis. J Cereb Blood Flow Metab 2011; 31:1443-1451.

Diffuse patterns of nonaneurysmal subarachnoid hemorrhage originating from the Basal cisterns have predictable vasospasm rates similar to aneurysmal subarachnoid hemorrhage.

Nonaneurysmal subarachnoid hemorrhage (SAH) has been historically associated with a benign clinical course. However, recent studies have suggested tha...
549KB Sizes 0 Downloads 13 Views