Radiologic imaging studies have become an integral diagnostic tool in the practice of clinical medicine. The role of body CT in particular has grown exponentially since its introduction in the mid 1970s. Today, more than 35 million CT scans are obtained annually in the United States alone. It is estimated that CT is now responsible for approximately 13% of all radiologic procedures performed in the United States and contributes 30% of the medical diagnostic radiation dosage to patients [1–3]. As the use of CT has markedly increased over the past two decades, so have its technical capabilities. CT now plays a critical role in the diagnosis and management of most patients with complex thoracic and cardiopulmonary disease, including pulmonary thromboembolism.

With the advent of multidetector scanners, CT has become the de facto gold standard for imaging pulmonary emboli [4]. Although the morbidity and mortality of unrecognized and untreated acute pulmonary embolism are well known, the potential carcinogenic effects of ionizing radiation on radiosensitive tissues such as the eye, thyroid gland, and female breast in particular are often not considered or deemed relevant to acute patient care. Because a large percentage of patients evaluated with CT pulmonary angiography are women of reproductive age, our objective was to estimate the effective radiation dose to the female breast incurred during these studies and compare that value with the average glandular breast dose for routine mammography. This information will allow referring physicians and radiologists to describe the effective radiation dose and relative risk of CT in terms patients understand. We also encourage the judicious use of CT pulmonary angiography and lower doses and nonionizing radiation alternatives when appropriate.

Our study involved four phases. First, we retrospectively reviewed the demographic data of all patients who underwent CT pulmonary angiography studies at our institution between May 2000 and December 2002. Second, the diagnostic outcome of each of these CT studies was reviewed and categorized as negative for pulmonary thromboembolism, nondiagnostic for establishing or excluding the diagnosis of pulmonary thromboembolism, and positive for pulmonary thromboembolism. Third, we divided the clinical status of studied patients into one of three categories: inpatient, outpatient, and emergency department. We did not include data on incidental pulmonary thromboembolism detected on scans acquired for other clinical concerns or diagnoses. Fourth, on the basis of our CT pulmonary angiography protocol and imaging parameters, our physicists estimated the effective radiation dose to the female breast of an average 60-kg female using the ImPACT CT (Impact Performance Assessment of CT, Bence Jones Offices, St. George's Hospital, London) dosimetry calculator and the CT dose index (CTDI) listed for Siemens Plus 4 scanners and compared that value with the average glandular breast dose for a typical two-view screening mammogram reported in the literature.

All CT pulmonary angiography studies were performed on Plus 4 or Volume Zoom units (Siemens Medical Solutions): 5-mm noncontrast images were acquired first from the thoracic inlet to the adrenal glands using 140 kV, 120 mAs, and a 1.5 pitch. The scanning time delay was determined by administering a 30-mL timing bolus of iohexol iodinated contrast material (Omnipaque 300, Amersham Health Products) with sequential stationary images acquired through the main pulmonary artery. Subsequently, 2-mm contrast-enhanced images reconstructed every 1.5 mm were acquired in the caudocranial plane after the IV administration of 170 mL of iohexol using 140 kV, 150 mAs, and a 2.0 pitch. Because our study straddled the transition to PACS, approximately 20% of the studies were reviewed both on hard copy and at workstations. The remaining 80% of studies were interpreted directly on the PACS monitors in the reading room for the thoracic imaging division. Fellowship-trained board-certified thoracic radiologists interpreted 90% of the CT studies. Board-certified nonthoracic radiologists interpreted the remaining studies.

During the defined study period, 1,325 CT pulmonary angiography studies were performed. As suspected, the majority of studies, 60% (797), were on female patients. Forty percent (528) of the studies were on male patients. Most of the scanned females were of reproductive age with active and radiosensitive fibroglandular breast tissue. The mean age of scanned females was 52.5 years (range, 15–93 years) (Fig. 1). One hundred ninetynine CT studies (25%) were performed on females less than 40 years old. One hundred forty-one examinations (17.7%) were on women 25–35 years old, 46 examinations (6%) on women who ranged in age from 20 years to less than 25 years, and 24 studies (3%) on those less than 20 years old.

The CT pulmonary angiography diagnostic outcomes in the female cohort included 401 studies (50.31%) negative for pulmonary thromboembolism, 151 studies (18.95%) nondiagnostic or inconclusive for pulmonary thromboembolism, and 245 studies (30.74%) positive for pulmonary thromboembolism (Fig. 2). Forty-one (29.1%) of the 141 CT studies performed on women 35 years old or younger were positive, 65 (46.1%) were negative, and 35 (24.8%) were nondiagnostic. Two hundred four (31.1%) of the 656 CT examinations performed on women older than 35 years were positive, 329 (50.2%) were negative, and 119 (18.14%) were nondiagnostic. Quantum mottle related to the patient's body habitus, respiratory motion degradation, and suboptimal vascular opacification was responsible for most nondiagnostic (4%) or inconclusive (15%) studies. The majority of these patients underwent no further diagnostic imaging and were managed empirically on the basis of the level of clinical suspicion.

The positive yield for CT pulmonary angiography studies was greatest for female inpatient and emergency department patients (Fig. 3). Sixty-four percent (157) of positive CT studies were in inpatient females, and 27% (66) of positive studies were in emergency department females. Only 7% (16) of positive studies were identified in outpatient females.

Between May 2000 and December 2002, 314 CT pulmonary angiography studies were performed in the emergency department. Two hundred thirteen scans were obtained on women, and 101 scans on men (female–male ratio, 2:1). The mean age of the female patients was 49 years (range, 19–90 years).

A total of 14 patients had more than one scan, including one individual who was scanned eight times, each time for suspected pulmonary thromboembolism disease. Ten (71%) of these 14 individuals were women. The mean age of this subgroup was 48.5 years (range, 30–64 years). Of the 314 scans, a D-dimer level was obtained in 89 patients (28.3%).

Using a D-dimer level of 500 ng/mL or less to exclude acute pulmonary thromboembolism, we found that 34 patients had negative D-dimer results. Twenty-four of these patients had a negative CT pulmonary angiogram and 10 patients had a positive CT examination. Fifty-five patients had a positive Ddimer (> 500 ng/mL). Thirty-six of these patients had a negative CT examination, and 11 had a positive CT examination. Eight of these scans (14.5%) were interpreted as inconclusive for acute pulmonary thromboembolism. The D-dimer level showed a sensitivity of 52.4%, specificity of 40%, a positive predictive value of 23.4%, and a negative predictive value of 70.6%.

Basic dosimetric quantities are absorbed dose, expressed in grays (1 Gy = 100 rad), and equivalent dose, expressed in sieverts (1 Sv = 100 rem). These quantities are numerically equal for diagnostic radiographs. Organ dose is expressed in milligrays or rads. The principal CT radiation dose descriptor, which integrates the radiation dose delivered both within and beyond the scanned volume, is the CT dose index (CTDI). Although the CTDI does not represent an actual radiation dose to a specific patient, it can be used as a standardized index of the average dose delivered from a CT scan and can be used to calculate the effective radiation dose [2, 3, 5]. The CTDI is calculated as follows:where T is the slice thickness, L is the length of the CT ionization chamber, D(z) is the dose as a function of the position along the z-axis of rotation, and dz is integration distance. The CTDI is determined for a centered singlescan dose profile and depends on the peak kilovoltage and milliampere-second settings and the distance from the focus to the center of rotation [3, 6]. Our physicists calculated the estimated female breast effective radiation dose using the ImPACT CT dosimetry calculator and the CTDI listed for Siemens Plus 4 scanners for the contrast-enhanced phase of our chest CT examinations. The calculated dose to the average 60-kg woman during CT pulmonary angiography was 2.0 rad (20 mGy) per breast. This dose compares with an average glandular breast dose of 0.300 rad (300 mrad or 3 mGy) for standard two-view screening mammography [7]. Our physicists did not calculate the radiation dose delivered to the breasts during the unenhanced image acquisition. Because unenhanced image acquisition is part of the scanning protocol and likewise contributes to the total breast dose, inclusion of this additional piece of data would have been insightful.

Our study confirmed our hypothesis that a large percentage of patients evaluated with CT pulmonary angiography are indeed women, 60% in our cohort, and that a significant percentage of scanned women (26.7%) are young, under 40 years old, with radiosensitive fibroglandular breast tissue. We also confirmed that the breast tissue is exposed to a much higher radiation dose, a minimum of 2.0 rad, than radiologists, medical physicists, and the American College of Radiology, according to their guidelines, would desire [2, 3, 7, 8].

The clinical diagnosis of pulmonary thromboembolic disease is both problematic and unreliable. It is difficult to triage patients for CT pulmonary angiography on the basis of clinical findings alone. Patient age is likewise unreliable for triage purposes. One might assume that the rate of pulmonary thromboembolism disease would be considerably lower in younger patients and that a patient's age might be used to triage the CT evaluation of patients with suspected pulmonary thromboembolism. Our study does not support that assumption: 31% of our female cohort who were younger than 35 years old had a positive CT pulmonary angiogram. Similarly, 31% of our female patients over 35 years had a positive CT pulmonary angiogram. Although this may be peculiar to our particular institution, we do not believe age alone is a reliable tool for triage, and we believe that women of any age with the clinical suspicion of pulmonary thromboembolism will continue to be evaluated with CT pulmonary angiography. In fact, the number of CT examinations is likely to continue to increase as the clinical threshold for scanning patients continues to decrease. Thus, it is imperative to reduce the exposure of breast tissue to ionizing radiation.

The issue of female breast radiation exposure during chest CT is relevant for many thoracic applications of CT, including CT pulmonary angiography. Exposures ranging from 2.0 rad, as in our study, to 5.0 rad, at the body surface, are not rare for conventional chest CT. CT studies that overlap scanned regions or rescan the same anatomic region of interest (e.g., unenhanced and enhanced scans) have 2–3 times the radiation dose of nonoverlapped scans. This would include the protocol currently used at our institution. Our physicists did not calculate this additional radiation dose delivered to the breasts during the noncontrast phase of CT pulmonary angiography, which also contributes to the total breast dose. Theoretically, this would have further increased our minimum dose estimate of 2.0 rad (20 mGy) by two- or threefold.

Dose rates for CT angiography are typically in the same range as conventional chest CT, 2.0–4.0 rad (20–40 mGy) [9]. This compares with an effective radiation dose equivalent of 0.06–0.25 rad (0.6–2.5 mGy) for twoview chest radiography and an average glandular breast dose of 0.300 rad (300 mrad or 3 mGy) for standard two-view screening mammography [8, 10]. Few physicians are aware that conventional diagnostic chest CT imparts a radiation dose of 2.0–5.0 rad (20–50 mGy) to the breasts of an average-sized woman. This dose is roughly equivalent to 10–25 two-view mammograms and up to as many as 100–400 chest radiographs [9–12].

The technical capabilities of CT have been further enhanced with MDCT. State-of-the-art MDCT provides exquisite anatomic detail, reduced examination time, and the ability to perform complex multiplanar vascular and 3D reconstructions that have had a dramatic impact on vascular imaging in particular. However, the radiation dose is 30–50% greater with 4-MDCT compared with singleslice helical scanners as a result of scan over-lap, positioning of the X-ray tube closer to the patient, and increased scatter created by the wider X-ray beam [3, 8, 12, 13].

The adverse biologic effects associated with radiation exposure are classified as either deterministic or stochastic. Deterministic effects result from cell death and are quantified by radiation dose to a particular region that has a threshold level. If this threshold is exceeded, cell death usually occurs. Such effects do not occur with the doses typically used for CT. The potential radiation risks in patients undergoing CT are due to stochastic effects. These effects are believed to result in carcinomas in the patient and genetic mutations and alterations in offspring of affected patients. The probability of stochastic effects depends on the amount of radiation absorbed. According to the International Commission on Radiation Protection (ICRP) Special Task Force Report 2000, the radiation doses used in CT often approach or exceed those levels known to increase the probability of nonfatal and fatal cancers [2, 14].

To date, no scientific study has reported a direct link between cancer and CT radiation. Most authors have drawn inferences of an increased risk of cancer on the basis of the outcomes of victims surviving the nuclear fallout in Hiroshima and Nagasaki. However, the increasing radiation exposure from CT and its potential adverse effects on both young patients and radiosensitive tissues is a real concern. Available data for radiation-induced cancers suggest that 1 mSv (0.1 rad) of radiation exposure may lead to five additional cancers in 100,000 exposed patients [15, 16]. Assuming a linear relationship between increasing radiation dose exposure and the stochastic effects of ionizing radiation on biologic tissue, one can extrapolate a possible additional 100 cancers per 100,000 exposed individuals from CT angiography for pulmonary thromboembolism evaluation alone. This falsely assumes all fibroglandular breast tissue has the same radiosensitivity, regardless of patient age, and breast volume.

The risk of diagnostic imaging procedures inducing breast carcinoma is not hypothetical. It has already been observed. A study of 1,030 women with scoliosis who routinely underwent multiple thoracic spine radiography examinations as young girls revealed a nearly twofold statistically significant increased risk for incident breast carcinoma [17]. Furthermore, radiologic technologists, exposed to low levels of occupational ionizing radiation, have elevated risks of female breast cancer, thyroid cancer, and melanoma [18].

Most physicians would agree that CT is an excellent imaging technique and that the benefit of any one given CT examination outweighs the risk of the study. Collectively however, the risk of several thousand or possibly million CT scans and not infrequently several in the same patient could become a public health issue. Recall, in our study, that 14 patients had more than one CT pulmonary angiogram, including one individual who was scanned eight times. Ten of these patients were young women with a mean age of 48.5 years. This same phenomenon is likely encountered at many other institutions across the country as well.

Efforts should be made to reduce the number of repeat scans and the CT radiation dose without impacting the clinical usefulness of the study. It is possible to reduce the radiation dose in performing chest CT examinations by modifying scanning protocols and technical settings. Radiologists typically do this by reducing the tube current (normally between 80 and 300 mAs, or mA/sec), increasing the table increment (pitch), reducing the exposure time and therefore the exposure level, and reducing the tube voltage [18, 19]. However, each manipulation is associated with a compromise in image quality and potentially in diagnostic information. For example, diagnostic images of the lung may be obtained with tube currents as low as 10–50 mA. However, this reduction in milliamperes is associated with an increase in noise and quantum mottle and may compromise the evaluation of mediastinal structures, lung findings, and the detection of distal segmental and subsegmental emboli and pulmonary nodules less than or equal to 5 mm in diameter [9, 19, 20].

On some MDCT scanners, increases in pitch are associated with a proportional increase in tube current to maintain similar noise. In such cases, the increase in pitch does not affect the radiation dose, but rather compromises image quality and the value of multiplanar reconstructions [21]. Reducing tube voltage below 110–120 kV leads to an unfavorable ratio of absorbed radiation and signal-to-noise ratio in areas of the body with high radiation absorption (e.g., shoulder girdle) and is not useful [9, 20, 21]. Radiologists and clinicians must be willing to sacrifice image quality and potentially diagnostic information if such manipulations in scanning parameters are to be the sole means of reducing patient radiation exposure.

Recently, manufacturers have responded to the need for radiation dose minimization and have implemented significant improvements in the newer, now widely available, 8- and 16-MDCT scanners [19]. One such innovation, similar to automatic exposure control in conventional diagnostic radiologic imaging, involves dynamic real-time tube current modulation based on body geometry and attenuation. Although the resultant radiation doses are lower than those incurred with the 4-MDCT scanners, the dose is still proportionally greater compared with the doses of other routine diagnostic imaging studies.

It is possible to reduce the radiation dose to superficial radiosensitive organs such as the female breast, thyroid gland, and eye during chest CT examinations without adversely affecting imaging quality or diagnostic information. Thin-layered bismuth radioprotective garments have been designed for this purpose (AttenuRad, Dyna Medical). Although available, these devices are not typically used outside of pediatric imaging centers. Bismuth garments are far more malleable to the body's surface, fitting better than traditional lead aprons or shields. Proper placement of these bismuth garments can reduce breast radiation exposure by 57%, thyroid gland dose by 60%, and that to the eye by 40% [22, 23]. The reusable protective breast shields cost approximately $165.00. The shields for the thyroid and eye are approximately $5.00 [22]. These shields could be used routinely, especially for female patients, whenever the head, chest, or abdomen is scanned. We have recently begun using the AttenuRad breast shield at our institution during CT pulmonary angiograms. We are also in the preliminary stages of developing our own custom-designed breast shield. We are interested in conducting a study with various-sized breast phantoms to evaluate the effectiveness of this breast shield on breast radiation dose reduction on our newly installed 16-MDCT scanners.

Alternative examinations, other than chest CT, are another means of reducing radiation exposure to radiosensitive organ systems. This may be feasible particularly in the female outpatient population who, according to our study, had positive CT pulmonary angiograms in only 7% of the cases. The poor diagnostic performance of the radionuclide ventilation–perfusion (V/Q) scans among patients with abnormal chest radiographs, significant cardiopulmonary disease, and emphysema is well known to clinicians and radiologists alike. However, in the setting of a normal chest radiograph and no history of significant cardiopulmonary disease, the V/Q scan can be effectively used to triage patients with suspected acute pulmonary thromboembolism [24]. The female breast receives less than 1 rad (< 1 cGy) of exposure during V/Q lung scanning [8]. Although CT is currently the preferred study, MR angiography is another alternative imaging technique that provides information similar to that obtained with CT and should be considered in those patients in whom iodinated contrast agents or cumulative ionizing radiation exposure is of concern (e.g., young patients, reproductive-age women, and especially women who have undergone multiple prior CT examinations). Contrast-enhanced 3D MR angiography for the evaluation of acute pulmonary thromboembolism has sensitivities ranging between 75% and 100% and specificities of 95–100%. The positive predictive value and negative predictive value are 87% and 100%, respectively, for central emboli. MR angiography does have limitations in its ability to detect subsegmental emboli and is also not appropriate for critically ill patients requiring close monitoring [24, 25].

The plasma D-dimer enzyme-linked immunosorbent assay (ELISA) has become recognized as a sensitive screening test for excluding acute pulmonary thromboembolism [26–29]. Numerous other qualitative and quantitative D-dimer assays have been introduced. One such quantitative assay, the immunoturbidimetric assay, has been shown to be equivalent to the ELISA [27]. D-dimer is a specific degradation product of cross-linked fibrin. The pairing of the D domains of fibrin monomers occurs only with full cross-linking of the monomers. D-dimer is thus elevated in the setting of deep venous thrombosis and pulmonary embolism. Unfortunately, disseminated intravascular co-agulation, pregnancy, sepsis, liver disease, trauma, surgery, and neoplastic disease are also associated with elevated D-dimer levels; thus, it cannot be used to triage such patients.

A positive D-dimer result is of limited value, especially in hospitalized or critically ill patients. However, a negative D-dimer result is potentially useful in excluding acute pulmonary thromboembolism, with reported negative predictive values of 91–100%—far exceeding that encountered in our study. Some advocate incorporating the D-dimer assay into the triage of emergency department and outpatient populations [26–29]. Our experience with the D-dimer assay has been unrewarding. Even though many of our patients had underlying neoplastic, rheumatic, and surgical diseases, which may explain the lower sensitivity (52.4%), specificity (40%), and negative predictive value (70.6%) than reported in the literature, we can understand the reluctance of our clinical colleagues to routinely integrate the D-dimer assay into their clinical algorithm. We are currently revisiting this issue at our institution.

The latency period for potential cancer induction is estimated to be 10–30 years in the dose ranges used in CT pulmonary angiography [8]. It is an accepted concept among radiation biologists and public health officials that the younger the patient is at the time of exposure, the greater their lifetime risk of developing nonfatal and fatal cancers. The greater lifetime risk is compounded by the increased biologic susceptibility to radiationinduced cancer. Thus, CT pulmonary angiography may not always be the best diagnostic option in young patients or reproductive-age and perimenopausal women. If CT is indeed justified in this patient population, every effort should be made to reduce the radiation dose, shield the patient, and limit the number of CT examinations performed.

In conclusion, the technologic advances in CT have had a dramatic impact on the diagnosis and management of almost all patients with complex thoracic and cardiopulmonary disease, including pulmonary thromboembolism. The role of CT will only continue to grow with the availability of MDCT and its multiplanar vascular and 3D reconstruction capabilities. However, the need for any given CT examination should always be justified on sound medical grounds. Both radiologists and referring clinicians need to work together to develop better selection criteria of patients for CT; reduce the radiation exposure of patients; and shield the eyes, thyroid gland, and female breasts whenever possible. In those situations in which the diagnostic yield of CT is expected to be low, alternative examinations should be considered.

Address correspondence to M. S. Parker ([email protected]).

Presented at the 2004 annual meeting of the American Roentgen Ray Society, Miami Beach, FL.