3.Introduction to Digital

Subtraction Angiography

Introduction to DigitalSubtraction Angiography

Digital subtraction angiography (DSA) is a newradiographic technology used in diagnosing vas-cular disease. DSA is employed to obtain imagesof arteries in various parts of the body and ishighly effective in contrasting arterial structureswith their surrounding bone and soft tissue (3).DSA has proven especially useful in the identifica-tion of vascular abnormalities, including occlu-sions, stenoses, ulcerated plaques, and aneurysms(21,58,107).

The potential importance of DSA in the diag-nosis of cerebrovascular disease is suggested byReuter’s (87) observation that as much as one-

TECHNOLOGICAL DEVELOPMENT

The development of DSA was a result of theresearch of medical physics groups at the Univer-sity of Wisconsin, the University of Arizona, andthe Kinderklinik in Kiel, West Germany duringthe early 1970s (21,58,74). Fundamental advancesin intravenous arteriography, which had been in-termittently used since the 1930s, were made pos-sible by the introduction of cesium iodine imageintensifiers and advances in digital electronicmethods of storing and manipulating information(21). By 1978, the feasibility of DSA for humansubjects was demonstrated, and prototype com-mercial DSA systems were introduced in 1980 atthe Universities of Arizona and Wisconsin, theCleveland Clinic, and South Bay Hospital inRedondo Beach, California (57,58,74). There are

quarter of the combined volume of neuroradiol-ogy and angiography services in some medicalcenters is now directed toward evaluating carotidand cerebral atherosclerosis, including stroke. TheCooperative Study of Transient Ischemic Attacks(TIAs) (102) reported an average of 5.4 definiteTIAs per 100 acute beds per year in the partici-pating medical centers. Estimates of the use ofarteriography procedures for these hospitalizedpatients range between 87 and 97 percent (23).DSA will either supplement or replace a large por-tion of the arteriographic procedures.

now nearly 20 manufacturers of DSA systems andmany more in the process of developing new sys-tems (18).

The size of the market for DSA equipment issomewhat difficult to estimate because of the un-certain future of demonstrated uses of DSA in cor-onary angiography. A spokesperson for one ofthe major manufacturers of DSA equipmentshared two projections of investment bankingfirms for all types of DSA units for the periodfrom 1982 through 1986. One firm projected totalsales of 5,160 units, while the other firm projectedsales of 9,800 units for the same period. Therewere estimated to be about 600 DSA units of alltypes in operational status as of January 1983.

CONTEMPORARY METHODS AND CLINICAL APPLICATIONS

With respect to cerebrovascular diagnostic in the evaluation of the extracranial circulation,studies, both intravenous and intra-arterial DSA especially the carotid arteries, in most cases.1

have been employed. However, in this case studythe notation “DSA” is used to signify intravenousapplications only. The focus is limited to intra- ‘As of October 1983, when this case study was submitted to OTAvenous DSA, because it is the method employed for final editing.

15

16

DSA systems work in the manner depicted infigure 3-1 as follows: a contrast medium is injectedintravenously; X-ray detection of the contrastmedium produces 1 to 30 exposures per second(before and after the injection of contrast me-dium); and arterial images are converted fromanalog to digital form and transmitted to a com-puter-storage complex (55). The digitalized im-age information makes it possible to “subtract”the precontrast images from those obtained aftercontrast injection so as to visualize arterial struc-tures without direct arterial puncture and injec-tion. The data can be recalled for viewing on avideo screen, and successive images createdthrough subtraction techniques which allow thecontrast of the arterial structures to be visualizedfor the detection of abnormalities.

The purpose of the subtraction process used inDSA is to eliminate (or factor out) the bone andsoft tissue images that would otherwise be super-imposed on the artery under study (12,58). Theserial images show changes in the contrast appear-ance over time (temporal subtraction) and at vary-ing X-ray intensities (energy subtraction) (12,57).

Most DSA examinations require 25 to 45 min-utes to perform (63,99,112), if there are no tech-nical complications (e.g., difficulties with catheter-ization), and can be performed on an outpatientbasis. This is a considerable advantage in safetyand cost over most standard arteriographic ex-aminations, which require at least overnightobservation of the patient in the hospital to detectpost-procedure arterial obstruction or hemorrhage(24,33). However, a small number of the latterhave been safely performed on an ambulatorybasis in recent years (45).

DSA has a wide range of clinical applicationsin addition to its use in carotid artery studies.Mistretta and his colleagues (74) at the Univer-sity of Wisconsin have substituted DSA for stand-ard arteriography in the evaluation of aortic archanomalies, aortic coarctation, and vascular by-pass grafts. Digital subtraction techniques havealso been used for imaging of the abdominal, car-diac, pulmonary, carotid, intracerebral, andperipheral vessels. Table 3-1 provides an overviewof the range of attempted applications of DSA im-aging technology reported in the literature. Be-

Figure 3-1 .—Diagram of a Digital Subtraction Angiography (DSA) System

II

III

/IIIIIIIIII

III

;IIII

SOURCE: Office of Technology Assessment. Adapted with permission of the General Electric Co, from G. S. Keyes, N. J, Pelc, S, J. Riederer, et al., Digita/ F/uorographyA Teclmology (@date (Milwaukee, Wl: General Electric Co., 1981),

17

Table 3-1 .—Attempted Applications of DSA Reported in the Literaturea

Anatomical regions studiedPrincipal author of study Carotid Thoracic Cardiac Abdominal Intracranial Pulmonary Peripheral Aorta Renal Other

Weinstein, et al. (1981)b. . . . . . X . . . . . . .Crummy, et al. (1980). . . . . . . . . X . . . . . . . . .Levy, et al. (1982). . . . . . . . . . . X . . . . . . . . .Brody, et al. (1982). . . . . . . . . . . X . . . . . . . . .Turnipseed (1982). . . . . . . . . . . . X . . . . . . . . .Buonocore, et al. (1981). . . . . . . . . . . . . . . . . .Mistretta, et al. (1981). . . . . X . . . . . . . . .Hillman, et al. (1981). . . . . . . . . . . . . . . . . . . .Johnson (1982) . . . . . . . . . . . . . . . . . . . . . . . . .Kruger, et al. (1981 )........ .. X.. . . . . . . .Meaney, et al. (1980 )....... .. X.. .. .X...Pond, et al. (1982 )....... . . . .. X.. . . . . . . .Chilcote. et al. (1981).. . . . . . .. X.. . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .x. . . . . . . . . . . . . . . . . . . . . . . . . . x. . . . ... .X.... . . . . . . . . . .x.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .aDoe~ not necess~ily imply routine clinical use at this timebFull cltatlons found In References Sect!on

SOURCE Off Ice of Technology Assessment

cause of problems with spatial and contrastresolution, virtually all examinations of the cor-onary arteries, which cannot be adequately vis-ualized with intravenous DSA at the present time,require arteriography.

Because DSA functions as both a screening pro-cedure for at-risk or asymptomatic patients andas an evaluative procedure for reconstructive sur-gery (74,108), estimates of the probable volumeof use based only on the latter type of use arebound to be conservative. DSA makes a uniquecontribution to the field of diagnostic radiology(3), serving as a bridging technique betweentotally noninvasive tests and conventional arteri-ography, at times replacing the latter. Turnipseedand his colleagues (107) clarified the respectiveuses of the techniques available to the radiologistand clinician for diagnostic imaging:

Arteriography has played an important rolein the surgical management of peripheral vas-cular disease because of its ability to preciselydefine the location and severity of arterial le-sions. However, its clinical use has been limitedby the risks of arterial catheterization, hospitali-zation cost, and poor patient acceptance. Arteri-ography is now commonly used to confirm adiagnosis of vascular disease and to plan appro-priate surgical management in patients withsymptoms and physical findings of arterial in-sufficiency.

Because arteriography has not been practicalfor routine diagnostic screening, a variety ofnoninvasive screening tests have been developed

as diagnostic aids. These noninvasive methodsallow more objective evaluation of larger patientpopulations and are attractive because of safety,cost efficiency, patient acceptance, and the ac-curacy in detecting hemodynamically significantocclusive lesions. These techniques have beenused primarily for diagnostic screening and post-operative assessment.

Although noninvasive methods are useful,they have limited capabilities and some seriousshortcomings. Many noninvasive tests are in-direct and restricted by technical limitations toevaluation of isolated arterial segments. Mostcannot define or distinguish minor stenosis andulceration from normal vessels and have diffi-culty in assessing remote areas of the circulation(intracranial, cardiac, and visceral systems).Noninvasive equipment is expensive, often veryspecialized, and requires personnel with specifictechnical and interpretive skills (107).

DSA, therefore, has the potential of significantlyimproving the radiologist’s and clinician’s capa-bilities in diagnostic screening and postoperativeassessment.

Furthermore, DSA is likely to limit the use ofolder noninvasive tests (e. g., periorbital ultra-sonography and thermography), because ofgreater sensitivity and specificity for diagnosticpurposes. However, the evaluation of DSA as adiagnostic technology for the study of carotidartery disease must note the increasing develop-ment and diffusion of ultrasound-based methodsfor diagnosing extracranial occlusive vascular dis-

18

ease. These technologies have developed very rap-idly and newer methods are likely in the future.Because of the safety of these tests, their popu-larity in clinical practice is not likely to bedisplaced by DSA in complex or poorly under-stood clinical situations. However, it may be ex-pected that, where the diagnosis of a carotid TIAseems highly likely on clinical grounds alone, thephysician may select DSA as the initial diagnos-tic test.

Of the large and growing number of nonin-vasive tests, ultrasound imaging has proved to bethe most versatile and reliable in clinical practice.A combination of B-mode real-time imaging, witha Doppler scanning device (often called “duplexscanning”), has become increasingly prevalent.Using this method, an image of the carotid vesselis obtained with the B-scan, and then the bloodflow pattern at a given anatomic location is deter-mined with the Doppler signal. This method isadvantageous in noninvasive diagnosis in skilledhands, but it takes considerable experience for anoperator to become sufficiently expert in the useof this tool to produce reliable and reproducibleinformation of clinical value. However, in partbecause these techniques have proved popularwith practicing clinicians, and in part because theyare affordable in office-based practice, industryis likely to respond to the demand for this tech-nology with more accurate and more easily per-formed duplex scanning in the near future.

By way of comparison, the indirect noninvasivetests (e.g., periorbital ultrasonography), whichmonitor the cerebral and orbital circulationsbeyond (downstream from) a carotid lesion, havebeen shown in most practice settings to have alower sensitivity and specificity than DSA and/orarteriography and are employed much less fre-quently at this time than was the case only a fewyears ago. This trend is likely to be accentuatedin the coming years.

As described above, DSA is also likely to limitthe use of, or substitute for, arteriography undermany clinical circumstances. In addition, it willbe employed in situations where arteriography isinapplicable, thus increasing the total volume ofarterial examinations. For example, some patientsfor whom arteriography is risky—such as elderly

patients, who are at greater risk for stroke fromcerebral arteriography--can have their carotidarteries examined easily and with reduced risk ofcomplication using DSA. Also, DSA may be per-formed repeatedly on the same individual in or-der to monitor postoperative or therapeutic prog-ress, without significant morbidity and with goodpatient compliance. The comparative advantagesof DSA and standard arteriography have beensummarized in table 3-2.

It is clear that DSA examinations are not a sim-ple substitute for arteriograms. Instead, for thoseconditions for which DSA and arteriography areboth applicable, the lower radiation and com-plication rates of DSA, the outpatient site oftesting, the reduced time required for the proce-dure, and patient acceptance (83,112) may resultin a use rate of DSA many times greater than thatof arteriography. In addition, DSA may be usedas a substitute for or a supplement to other nonin-vasive tests. Two studies of DSA cost effective-ness estimate that for patients with suspectedTIAs, clinicians order DSA examinations (wherethis technology is available) at approximatelytwice the current rate of arteriographic studies(24,33). However, at present there are no em-pirical bases for future estimates of DSA utili-zation.

Table 3.2.—Comparative Advantages of DSAand Conventional Arteriography

Advantages of standardAdvantages of DSA arteriography

Decreased morbidity Increased spatial resolutiona

Decreased patient Feasibility of selectivediscomfort injections

Decreased hospitalization Less degradation of patienttime motion

Decreased procedure time Visualization of smallerDecreased film cost blood vesselsIncreased contrast

resolution b

Usefulness in patientswith limited arterialaccess

Lower cost perexamination

asPafjal ~e~olut~on: extent to which radiographic image makes it possible todetect and distinguish anatomically contiguous structures.

bcontrast reso/u//on: extent to which computer can detect subtle differencesin amount of contrast medium present

SOURCE: M. F. Steighorst, C. M. Strother, C. A. Mistretta, et al., “Digital Sub-t rac t ion Arrgiography: A Clinical Overview, ” Applied F?adlo/ogy10(6):45-49, 1931.

19

CLINICAL EFFICACY AND EFFECTIVENESS OF DSAThe claimed advantages and disadvantages of

DSA derive, in large part, from the efficacy andsafety of the technology. A substantial volumeof evidence regarding the efficacy and safety ofDSA is now available through clinical testing ofDSA in several medical centers.

Banta and Behney (8) define technological ef-ficacy as “the probability of benefit to individualsin a defined population from a medical technol-ogy applied for a given medical problem underideal conditions of use” (emphasis added). Effec-tiveness is the probability of benefit under aver-age conditions of use. The literature to date onDSA generally addresses clinical efficacy andsafety, not effectiveness. Most studies have beenconducted in institutions engaged in clinical re-search under carefully monitored conditions (21).An exception is the experience documented atScottsdale Memorial Hospital in Arizona (63). Itis not clear whether experimental and early clini-cal data from academic medical centers, such asthe Universities of Wisconsin and Arizona—eachwith several years of pioneering experience inDSA use—can be employed reliably to predict theeffectiveness and safety of DSA by radiologistsand clinicians in community hospitals, clinics, andgroup practices and the resulting patterns of DSAuse.

Measurement of the efficacy of DSA is multi-dimensional, as depicted in table 3-3 (adapted

from Fryback [38]), because benefits can bediscerned at the levels of: 1) physical image; 2)the detection, accuracy, and sensitivity of tests;3) diagnostic decisionmaking; 4) therapeutic deci-sionmaking (or “management efficacy”); 5) pa-tient outcome; and 6) social utility (38). Implicitin this scheme is the belief that increasing diag-nostic accuracy is not an end in itself, but ratheran instrumental value. The overall efficacy ofDSA, then, lies in its contributions to better pa-tient outcomes and ultimately to improved socialwelfare (31,38),

Most of the clinical evaluations of DSA havetaken place at levels 1 and 2 of the Fryback model,namely, with a focus on image quality or on diag-nostic sensitivity, specificity, and accuracy. “Sen-sitivity” may be defined as the proportion ofpositive tests in all patients with disease; “speci-ficity” is the proportion of all negative tests in pa-tients without disease; and “accuracy” is the ratioof correct diagnoses to all diagnoses.

Investigations of the efficacy of DSA have notyet concentrated on the effects of DSA through-out the medical care system. The literature gen-erally does not address differences in physiciandiagnosis, selection of treatment alternatives, pa-tient outcomes, or social welfare attributable toDSA. The majority of studies consider the ac-curacy of DSA for diagnosis in comparison toother diagnostic techniques, usually conventional

Table 3-3.—Levels At Which Diagnostic Technologies May Be Assessed

Level of the measurement Typical output measures

Level 1: Image efficacy . . . . . . . . . . .

Level 2: Image andobserver efficacy . . . . . . . . . . . . . . .

Level 3: Diagnostic efficacy . . . . . . .

Level 4: Management efficacy(therapeutic decisionmaking) . . . .

Level 5: Patient outcome efficacy . .

Level 6: Societal efficacy

quality of image resolution

percentage yield of abnormal cases; percentagecorrect diagnoses; sensitivity; specificity

change in order of clinician’s diagnosticconsiderations

percentage change in therapeutic protocol;percentage change to appropriate therapy

survival rates; percentage cures; morbidity measures;reduced worry of patient and family

(or utility) . . . . . . . . . .’. . . . . . . . . . . dollars added to GNP; age-adjusted survival ratesSOURCE: Off Ice of Technology Assessment, Adapted from D G, Fry back, “A Conceptual Model for Output Measures in Cost.

Effectiveness Evaluation of Diagnostic Imaging, ” paper presented at the Symposium International de EvaluationCout. Eff!caclte en Neuroradiologie, Bordeaux, France, May 14.15, 1982.

20

arteriography. For images of good diagnostic ing the scan procedure. The quality of images canquality (approximately 85 percent), the sensitivity be affected by swallowing, breathing, peristalsis(95 percent), specificity (99 percent), and accuracy (movement of esophagus), or other physical mo-(97 percent) of DSA appear to be very good tions, depending on the vessels to be visualized(19,112). (112). Other causes of failure are inadequate ve-

in general, the principal reason for inadequatenous access, leakage of dye, and faulty injection.

visual resolution is movement of the patient dur-

THE SAFETY (ASSOCIATED PATlENT RISKS) OF DSA

OTA defines the “safety” of a medical technol-ogy to be:

. . . the judgment of the acceptability of relativerisk in a specified situation;

while “risk” is defined as:

. . . the probability of an adverse or untowardoutcome occurring and the severity of the resul-tant harm to health of individuals in a definedpopulation associated with use of a medical tech-nology applied for a given medical problemunder specified conditions for use (8).

Using these definitions, the safety of DSA maybe measured and compared to the present meth-ods of diagnostic imaging for vascular diseases.As noted by Patterson (81), the efficacy and safetyof neurosurgical and related technologies are onlyestimated informally, because of measurement dif-ficulties, different conditions of use, and the ex-perimental nature of some technologies. This ap-pears to be the case for DSA. There are norigorous epidemiological or randomized clinicalstudies that document either the direct or indirectsafety effects. This seems to be due to the fact that

most clinicians are impressed with the apparentlow risk of DSA relative to benefit, as comparedwith alternatives.

The amount of radiation from DSA is so smallthat clinical decisionmaking generally does nottake radiation into consideration. Radiation dosesfor DSA reported in carotid studies are given intable 3-4. Radiation exposure varies depending onthe subtraction method selected for the exam andthe number of views required for diagnosis (21,24,56).

Complications from DSA are minimal com-pared to standard arteriography. Both peripheraland central intravenous injections are likely to beless risky than arterial punctures required forarteriography. In particular, arteriographic pro-cedures may cause stroke due to dislodging em-bolic material (19), dissecting the arterial walls,or rupturing aneurysms (54), whereas this doesnot occur with DSA.

A survey of radiologists at 514 hospitals showedthat about 2 to 3 percent of all patients undergo-

Table 3-4.—DSA Radiation Exposure Estimates for Cerebrovascular Studies

Estimated radiationPrincipal author of study Area of study exposure per imagea

Crummy, et al. (1980)b Extracranial arteries 100-150 mRc

Chilcote, et al. (1981) Carotid arteries 230 mRBrody, et al. (1982) Carotid arteries 130 mRDetmer, et al. (1982) Extracranial arteries 3.6 Rd

aNurnber of images per study may vary considerably depending on anatomical site, operators, or Patient characteristics.bFull citations found in References Section.cmR . milliroentgen (I Roentgen (R) = the unit of radiation exposure equivalent to 2.08 x 10° iOfI pairs in one cc. of air. The

measure of radiation exposure in human tissue conventionally used is the “Rad, ” the absorbed dose of 100 ergs/gin of tis.sue. One Rad is equivalent to 1.02 Roentgen, the difference between two units of measure being considered negligible,)

dR . Rad. (Detmer, et ai, (24), the source for data on a complete DSA study (including multiple ima9eS), note that a convention-al arteriographic study would expose the patient to approximately 20 Rads of radiation.)

SOURCE: Office of Technology Assessment,

.

21

ing a total of 118,591 arteriographic exams—transfemoral, transaxillary, translumbar—suf-fered complications which required additionaltherapy or prolonged the patient’s hospital stay(54). These complications included 30 deaths (0.03percent of all exams). Another study indicatedthat of 1,328 patients who were suspected of hav-ing TIAs and had arteriograms, 13 percent hadtemporary complications and 0.65 percent suf-fered permanent neurological complications (102).Johnson (56) concludes that:

“(t)he complexity, expense, and a certain mor-bidity and mortality associated with this radio-graphic procedure [arteriography] compel a setof indications virtually as strict as that for sur-gery.”

Certain complications can arise in DSA examsdue to leakage of contrast medium outside thevein, venous reflux (contrast medium going intovein the wrong direction), or patient reaction to

the contrast medium (21,57,108,112). Variousclinical studies have documented only a smallnumber of such problems—all were transient—of several thousand patients examined (19,21,57,112).

The safety as well as the clinical efficacy of DSAwill depend to a considerable extent on the qualityof the particular equipment being used (a factorwhich also affects diagnostic accuracy and speci-ficity); the compatibility of new DSA equipmentwith existing facilities; the techniques used; andthe experience of the physicians and allied healthpersonnel performing and interpreting these diag-nostic images. Several investigators have notedsome variability in imaging capabilities of DSAunder different technical conditions and haveevaluated the physical requirements of the imag-ing systems with regard to assuring high qualitystandards (6,67).