586 Hemorrhagic InfarctsROBERT G. HART, M.D., AND J. DONALD EASTON, M.D.

MORE THAN A HUNDRED YEARS AGO, JohnLidell observed that "red softening commences after24—48 hours, dating from the apoplectiform attack" ofcerebral embolism.1 While his pathological descrip-tions are arguably ambiguous, Lidell's observationssurely are among the earliest descriptions of hemor-rhagic infarct (HI). Our clinical-pathological conceptsof His then lay relatively dormant until 1951, whenFisher and Adams emphasized the association of HIand embolism and proposed their well-known hypoth-esis about the mechanism of hemorrhagic transforma-tion.2 HI remained an autopsy diagnosis until comput-ed tomography (CT) and magnetic resonance imaging(MRI) permitted diagnosis during life. Recent studies,including two in this issue of STROKE,3' * have refinedour clinical concepts of HI.5"10 Treatment of acutestroke patients with thrombolytic agents and antico-agulants, both having secondary brain hemorrhage astheir most feared complication, has rekindled interestin the occurrence and mechanisms of hemorrhagic in-farction."15

Hemorrhagic infarction describes multifocal, secon-dary bleeding into brain infarcts. Innumerable foci ofcapillary and venular extravasation either remain asdiscrete petechiae or merge to form confluent purpura(fig. 1). Pathologically, the distinction between paleand hemorrhagic infarcts is arbitrary in mild cases.Most recent infarcts show a few scattered petechiae,along their margins. Most pathologists do not desig-nate an infarct as hemorrhagic unless a large part isinvolved with petechiae or unless petechiae are conflu-ent.2- 17

Autopsy studies report that about 30% (range18-42%) of recent brain infarcts are hemorrhagic, withthe wide range in prevalence largely explained byvarying definitions of mild HI and by ill-defined inter-vals from stroke to death.2'3- "-20 Autopsy studies em-phasize a high prevalence of HI associated with embol-ic stroke: His occur in 51-71% of "embolic" strokescompared to 2-21% of "nonembolic" strokes.2'3i " Inmost cases, the inability to diagnose embolic strokewith certainty, even at autopsy, clouds the interpreta-tion of both autopsy and clinical data.

CT and autopsy studies convincingly link HI withinfarct size.3 V7> l4'2I Lodder and colleagues in this is-sue question whether the association of HI and car-dioembolic stroke at autopsy is explained by infarctsize rather than by the stroke mechanism.3 If cardioem-bolic strokes are unduely large, HI could be a markerof large infarcts rather than the stroke mechanism perse. Earlier autopsy studies of HI did not report the

From the Department of Medicine (Neurology), University of TexasHealth Science Center, 7703 Floyd Curl Drive, San Antonio, Texas,USA 78284.

Address correspondence to Drs. Hart& Easton, Department of Medi-cine (Neurology), University of Texas Health Science Center, FloydCurl Drive, San Antonio, Texas 78284.

Received May 14, 1986; accepted May 21, 1986.

interval from stroke to death, the cause of death, or thecorrelation of infarct size with HI, thus confoundinganalysis of the relationship of HI to infarct size inde-pendent of stroke mechanism.2'17'l9 Using brain herni-ation to reflect infarct size, Lodder and colleaguesconclude that "in the cases dying from brain herni-ation, HI was equally frequent in the group with acardiac embolic cause of stroke, as it was in the groupwithout such cause."3 However, alternative interpreta-tion of these data yields a different conclusion. Wewould not consider the few petechiae termed "smallHI" by Lodder and colleagues (their fig. 2, indicatedby arrows) as HI. Considering only confluent pete-chiae (the unequivocal HI in their fig. 1), His werepresent in 42% (6 of 14) patients with cardioembolicsources who herniated compared with 15% (2 of 13) ofpatients without cardioembolic sources who herniated.It is our view that these important data of Lodder et alsupport the notion that cardioembolic strokes have aspecial, but certainly not exclusive, propensity forhemorrhagic transformation. Relatively dense His thatoccur deep within the distribution of the affected vesselare most often associated with a cardioembolic mecha-nism.

His are observed on less than 5% of initial CTs ofpatients with acute ischemic stroke7-n but in an addi-tional 10% if CT is repeated in the subsequentweeks.6'823"23 Hemorrhagic transformation visible onCT is delayed for variable intervals following stroke;early hemorrhagic transformation (within 48 hours ofstroke onset) appears to be particularly frequent withpresumed cardioembolic strokes.4'6-8-92* Horning etal performed serial CTs over a four week interval in 65patients with acute ischemic stroke and reported hem-orrhagic transformation in 17% within the first weekafter stroke, an additional 23% in the second week, 3%in the third week, and none in the fourth week.6 Theoverall 43% prevalence of HI is considerably higherthan other clinical series of serial CTs8 and even ex-ceeds the autopsy prevalence, suggesting patient popu-lation differences or potentiation of HI by their use ofosmotic agents and low-dose heparin.6

In one small series, enhancement of acute (within 28hours) infarcts on CTs performed three hours afterhigh-dose intravenous contrast infusion was predictiveof later development of HI.23 HI are readily detected byMRI, but systematic comparison with CT has not yetbeen undertaken. Surprisingly, HI has not been associ-ated with acute or chronic hypertension in most clinicaland autopsy studies.3-'•l7-18-27

Given the dynamic nature of secondary hemorrhagictransformations and their relationship to large infarcts,it is difficult to directly compare prevalences of HIbetween CT and autopsy studies. Large infarcts wouldbe over-represented in autopsy studies of recent strokeand postmortem examination would detect HI withgreater sensitivity than CT; hence HI should occurmore frequently in autopsy studies. On the other hand,

Dow

nloaded from http://ahajournals.org by on O

ctober 12, 2021

HEMORRHAGIC INFARCTS/tfa/7 and Easton 587

FIGURE 1. Variable degrees of hemorrhagic infarction. A) Scattered petachaie in a "watershed" infarct. B) Dense, confluentpurpura in a cardioembolic stroke (from Escourille and Poirier, with permission).16 Different degrees/patterns may haveimplications for stroke mechanism.

late-developing HI may especially occur in stroke sur-vivors and result in higher CT detection rates.

About half (range 29-73%) of all His in clinical-CTstudies are associated with presumed cardioembolicstrokes.4'6'7i 10 '" Of patients with presumed acute car-dioembolic stroke, HI will be present on only 5% ofCTs performed within the first 24 hours6' M- a - N but inabout 20% (range 8-62%) of CTs performed 1-2weeks after s t roke . 6 1 0 1 4 2 4 3 0 Hemorrhagic transfor-mation is delayed for 6-12 hours after cardioembolicstroke, but usually occurs within 48 hours in nonanti-coagulated patients.4931 This relatively early hemor-rhagic transformation (within 48 hours) associatedwith presumed cardioembolic strokes compared tonon-embolic infarcts partly explains why studies ofearly CT or of autopsies with early stroke deaths reporta particularly high association of HI with cardioem-bolic stroke.

Serial CT studies of HI in patients with presumedcardioembolic stroke have yielded two additionalclinical correlations: (1) most hemorrhagic transforma-tions in nonanticoagulated patients are not associatedwith clinical worsening and (2) anticoagulation regu-larly accentuates the degree of HI, often with clinicaldeterioration. In one large collection of nonanticoagu-lated patients with presumed cardioembolic stroke andevolving HI documented by serial CTs, only 17% (2 of12) experienced clinical deterioration associated withhemorrhagic transformation.5'9 Hemorrhagic transfor-mation never occurred in the initial six hours afterstroke onset, but the majority were present between24—48 hours.31 Among anticoagulated patients withhemorrhagic transformation documented by serial CT,56% (15 of 27) deteriorated.9 Severe secondary hemor-rhage, resembling confluent hematoma, occurred in41% (11 of 27) of anticoagulated patients but only 8%(1 of 12) of nonanticoagulated patients.9 Other clinicaland autopsy studies have confirmed that anticoagula-tion during the period of hemorrhagic transformation

accentuates the degree of secondary hemorrhage, oftenwith clinical worsening, in patients with presumed car-dioembolic stroke.4'17 Anticoagulation does not appearto increase the likelihood of hemorrhagic infarction,but appears to accentuate the degree of hemor-rhage.14'26 The period of active bleeding is transientand anticoagulants administered after the appearanceof HI on CT do not usually result in clinical or radio-graphic worsening.424

What mechanism(s) allows blood to leak into someinfarcts, but not into others? In mild-moderate His,hemorrhage is believed to result from diapedesisthrough ischemic endothelium, usually without vesselrupture. An ischemic insult of sufficient degree to in-duce a transient abnormality in vascular permeabilityappears to be a necessary substrate for HI. The extentof disruption of endothelial junctions and capillaryrupture are related to the duration and degree of ische-mia in primates with transient middle cerebral arteryocclusion.32 A second essential condition is postulatedto be restoration of circulation to the injured capillarybed during the interval of increased permeability, ei-ther by re-opening of the initial site of occlusion or byestablishment of collateral circulation.2'l9

In other patients, variable degrees of necrosis of thevessel walls are present and the severe secondary hem-orrhage appears unifocal, with mass effect and intra-ventricular extension.3' '• "•33 The bleeding in these pa-tients clinically resembles intracerebral hemorrhagefrom a single focus, and some clinicians have suggest-ed two distinct syndromes of hemorrhagic transforma-tion: (1) multifocal hemorrhagic infarction and (2) sec-ondary hematoma (apparently unifocal) within aninfarct.3'6- "•M We suspect that, pathologically andmechanistically, the difference is of degree rather thanof kind: Both are points along a spectrum of vascularinjury, transiently increased permeability and secon-dary hemorrhage.

Hypothetically, the propensity of large infarcts to

Dow

nloaded from http://ahajournals.org by on O

ctober 12, 2021

STROKE VOL 17, No 4, JULY-AUGUST 1986

undergo hemorrhagic transformation may be a reflec-tion of the probability that the ischemia is sufficient toalter vascular permeability. In watershed infarcts,which may be large in volume and in which earlyreperfusion would be expected to occur, ischemia maybe insufficient to substantially alter vascular integrityand dense HI is uncommon.1735 HI is present in aboutone-third of patients with brain infarction due to vaso-spasm secondary to subarachnoid hemorrhage, withonset during remission of vasospasm.36

In cardioembolic infarcts, hemorrhagic transforma-tion is postulated to occur when distal migration of theembolic fragment from its initial site of arterial ob-struction allows reperfusion of the damaged vascularbed (fig. 2).2' l7~19 The relationship between re-openingof the site of occlusion and hemorrhagic transforma-tion is strongly supported by both radiographic andpathologic data.2'IOi l7> l8'38> 39 However, persistentproximal occlusions have been found in 11-17% ofHis associated with cardioembolic stroke, implicat-ing other routes of reperfusion as well.10'17' '9 The roleof collateral circulation-reperfusion in hemorrhagictransformation has not been thoroughly studied exceptin experimental models.

Do strokes resulting from artery-to-artery embolismhave the same predisposition for and mechanism of HIas those due to cardiogenic emboli? There are fewclinical or pathologic data addressing this issue, due tothe difficulty of defining this mechanism with certain-ty. We speculate that artery-to-artery embolism un-commonly results in sufficient ischemia to derangevascular permeability. However, with particularlylarge artery-to-artery emboli, the situation may well beanalogous to cardioembolic stroke.4

From analysis of these data, including those present-ed in this issue of STROKE, we conclude that car-dioembolic strokes do indeed have a special propensityfor hemorrhagic transformation, independent of infarctsize. Further, the early re-opening due to distal migra-tion of embolic fragments appears to result in a recog-nizable pattern of early (within 48 hours) hemorrhagictransformation with central or globular and relativelydense HI (fig. 3).2> 8 Development of collateral circula-

Brain surface

Pressure restored

\ Hemorrhagic infarction

Anemic InfarctionEmbolism breaks up.goes into one branch

FIGURE 2. Hemorrhagic transformation in embolic stroke.Re-perfusion during the initial days due to distal migration ofemboli from the initial site of occlusion may be a key factor inhemorrhagic transformation. From Toole, CerebrovascularDisorders, with permission.37

FIGURE 3. Subcortical hemorrhagic transformation may beparticularly associated with presumed cardioembolic stroke.None of these four patients with presumed cardioembolic strokereceived anticoagulation, one (upper right) deteriorated associ-ated with hemorrhagic transformation. From the case collec-tion of the Cerebral Embolism Study Group.9- 21

tion may be important in late hemorrhagic transforma-tion (after one week); these His tend to be scattered orgyral in pattern, near the infarct periphery and havelittle predictive value for stroke mechanism. Thesegeneralizations are not absolute. Many exceptions areknown and may be explained by variations in the tim-ing of re-opening of embolic occlusions and of devel-opment of the collateral circulation.

It appears that the mechanism of secondary hemor-rhagic transformation is a remarkably complex anddynamic process, involving a combination of vascularinjury with altered permeability and reperfusion, inte-grated over time. Standing on the shoulders of giants' •2

with the aid of CT and MRI imaging, clinicians nowhave new concepts about HI to consider, assimilateand further define.

References1. Lidell JA. A Treatise on Apoplexy. New York: William Wood and

Co, p. 193-195, 18732. Fisher CM, Adams RD. Observations on brain embolism with

special reference to the mechanism of hemorrhagic infarction. JNeuropath Exp Neurol 10: 92-93, 1951

3. Lodder J, Krijne-Kubat B, Broekman J. Cerebral hemorrhagicinfarction at autopsy: Cardiac embolic cause and the relationship tothe cause of death. Stroke 17: 1986 (this issue)

4. Ott BR, Zamani A, Kleefield J, Funkenstein HH. The clinical

Dow

nloaded from http://ahajournals.org by on O

ctober 12, 2021

HEMORRHAGIC WFARCTS/Hart and Easton 589

spectrum of hemorrhagic infarction. Stroke 17: 1986 (this issue)5. Cerebral Embolism Study Group. Immediate anticoagulation of

embolic stroke: Brain hemorrhage and management options.Stroke 15: 779-789, 1984

6. Horning CR, Domdorf W, Agnoli AL. Hemorrhagic cerebral in-farction — A prospective study. Stroke 17: 179-185, 1986

7. Lodder J. CT-detected hemorrhagic infarction; relation with thesize of the infarct, and the presence of midline shift. Acta NeurolScand 70: 329-335, 1984

8. Seto H, Nonaka N, Kuratsu J, Ito Y, Miura G, Matsukado Y.Clinical features of hemorrhagic infarction. Neurol Med Chir (To-kyo) 24: 706-711, 1984 (English abstract)

9. Cerebral Embolism Study Group. Proceedings of the Third Interna-tional Brain-Heart Conference (Cardioembolic Stroke Workshop)(Trier, Sept 1985). Martin Ninjhoff (Amsterdam), 1986

10. Yamaguchi T, Minematsu K, Choki Jl, Ikeda M. Clinical andneuroradiological analysis of thrombotic and embolic cerebralfunction. Jap Circ J 48: 50-58, 1984

11. del Zoppo GJ, Zeumer H, Harker LA. Thrombolytic therapy instroke. Stroke 17: 1986 (this issue)

12. del Zoppo GJ, Copeland BR, Waltz TA et al. The beneficial effectof intracarotid urokinase on acute stroke in a baboon model. Stroke17: 1986 (this issue)

13. Duke RJ, Turpie AG, Bloch RF, Trebilcock RG. A randomizedtrial of heparin in acute partial thrombotic stroke. Stroke 17: 133,1986 (abstract)

14. Cerebral Embolism Study Group. Immediate anticoagulation ofembolic stroke: A randomized trial. Stroke 14: 668-676, 1983

15. Bogousslavsky J, Regli F. Anticoagulant-induced intracerebralbleeding in brain ischemia. Acta Neurol Scanda 71:464-471, 1985

16. Escourille R, Poirer J. Manual of Basic Neuropathology. (SecondEdition) W.B. Saunders (Philadelphia) 1978, p. 85

17. Jorgensen L, Torvik A. Ischemic cerebrovascular disease in anautopsy series. Part 2. J Neurol Sci 9: 285-320, 1969

18. Uemura K, Okudera T, Ishii K, Fukasawa H. Hemorrhagic cere-bral infarction; its neuroradiological findings and pathogenesis. JapHeart J 22: 29-31, 1977

19. Adams RD, Van der Ecken HM. Vascular diseases of the brain.Ann Rev Med 4: 213-252, 1953

20. Britton M, Gustafsson C. Nonrheumatic atrial fibrillation as a riskfactor for stroke. Stroke 16: 182-187, 1985

21. Matias-Guiu J, Alvarez J, Davalos A, Codina A. Heparin therapyfor stroke. Neurology 34: 1619-1620, 1984

22. Cronquist S, Brismar J, Kjelin K, Soderstrom CE Computer as-sisted axial tomography in cerebrovascular lesions. Acta Radiol,16: 135-144, 1975

23. Fisher M, Zito JL, Silva A, DeGirolami U. Hemorrhagic infarc-tion: A clinical and CT study. Stroke 15: 192, 1984 (abstract)

24. Hakim AM, Ryder-Cooke A, Melanson D. Sequential computer-ized tomographic appearance of strokes. Stroke 14: 893-895, 1983

25. Hayman LA, Evans RA, Bastian FO, Hinck VC. Delayed high-dose contrast CT: Identifying patients at risk of massive hemor-rhagic infarction. Am J Radiol 136: 1151-1159, 1981

26. Calandre L, Ortega JF, Bermejo F. Anticoagulation and hemor-rhagic infarction in cerebral embolism secondary to rheumatic heartdisease. Arch Neurol 41: 1152-1154, 1984

27. Cerebral Embolism Study Group: Brain hemorrhage and cardioem-bolic stroke: Clinical observations, (submitted)

28. Furlan AJ, Cavalier SJ, Hobbs RE, Weinstein MA, Modic MI.Hemorrhage and anticoagulation after nonseptic embolic brain in-farction. Neurology 32: 280-282, 1982

29. Martin GJ, Biller J. Nonseptic cerebral emboli of cardiac origin.Arch Int Med 144: 1997-1999, 1984

30. Weisberg LA. Nonseptic cardiogenic cerebral embolic stroke:Clinical-CT correlations. Neurology 35: 896-898, 1985

31. Cerebral Embolism Study Group. Timing of hemorrhagic transfor-mation of cardioembolic stroke. Stroke 17: 558

32. Garcia JH, Lowry SL, BriggsL, etal. Brain capillaries expand andrupture in areas of ischemia and reperfusion. In Reivich M, HurtingHI, Cerebrovascular Diseases. Raven Press, New York, 1983, p.169-180

33. Bass E. Anticoagulation in cerebral embolism. Canadian J NeurolSci 10: 32-36, 1983

34. Bogousslavsky J. Anticoagulation and bleeding into embolic in-farcts. Arch Neurol 42: 1033-1034, 1985

35. Torvik A. The pathogenesis of watershed infarcts in the brain.Stroke 15: 221-223, 1984

36. Terada T, Komai N, Hayashi S et al. Hemorrhagic infarction aftervasospasm due to ruptured cerebral aneurysm. Neurosurgery 18:415-418, 1986

37. Toole JF. Cerebrovascular Disorders. Raven Press (New York),1984; p 187-198

38. Irino T, Taneda M, Minami T. Sanguinaeous cerebrospinal fluid inrecanalized cerebral infarction. Stroke 8: 22-24, 1977

39. Dalai PM, Shah PM, Aiyar RR. Arteriographic study of cerebralembolism. Lancet 2: 358-361, 1965

Dow

nloaded from http://ahajournals.org by on O

ctober 12, 2021