In countries with developed medical infrastructure, the use of breast fine-needle aspiration biopsy (FNAB) cytology has had its share of challenges over the past 20 years, among them the use of core needle biopsies. In developing countries where the use of FNAB cytology has been increasing rapidly, breast lesions are one of the most common sites sampled by FNAB. In 2016, the International Academy of Cytology Executive Council put together a “Breast Group,” which consists of cytopathologists, surgical pathologists, radiologists, surgeons, and oncologists working in breast care, with the aim of producing a comprehensive and standardized approach to breast FNAB cytology reporting.1,2
This approach will address the current challenges to FNAB cytology and include best-practice guidelines for the indications for breast FNAB cytology and the techniques of FNAB, smear making, and material handling. It will include a practical, standardized reporting system, including report content requirements, with defined descriptive terms and categories, structured reports with checklists and formats, and recommendations for the use of ancillary diagnostic and prognostic tests and suggested management algorithms. A standardized approach with best-practice guidelines will improve FNAB and smear-making technique, training, routine reporting, and quality assurance programs. If linked to management recommendations, it will improve clinicians’ understanding and use of FNAB cytology services.
Why a new reporting system?
Since the 1996 National Cancer Institute consensus meeting made recommendations for breast FNAB cytology reporting, there have been many developments in the diagnostic workup of breast lesions in surgeons’ rooms, breast clinics, and mammographic screening program assessment clinics, including the use of tomographic mammography, ultrasound, and MRI.3 There have also been significant developments in the role of various diagnostic procedures in management algorithms, and the use of breast FNAB cytology now varies greatly between breast clinics for symptomatic women, mammographic screening program assessment clinics, and hospitals in various cities and states as well as between developed and developing countries.
Breast FNAB cytology does offer many advantages because it is quick, is minimally invasive, causes minimal physical and psychological discomfort, and is acceptable to patients. It is a relatively inexpensive test. It enables rapid onsite evaluation (ROSE) and provisional reporting, which is ideal for multidisciplinary “one stop” diagnostic clinics that provide same-day clinical, radiological, and provisional cytological assessment.
Breast FNAB is a highly specific and sensitive test to accurately diagnose benign and malignant lesions when undertaken by an operator experienced in the biopsy technique and cytopathologists experienced in reporting breast cytology. It is cost-effective for the preoperative diagnosis of palpable and ultrasound-detected impalpable breast lesions. It can also provide formalin-fixed, paraffin-embedded cell blocks for immunohistochemistry for prognostic indicators, including estrogen and progesterone receptors and for in situ hybridization for HER2. Material from directly smeared slides or cell block material can be used for PCR and other potential molecular testing.4
In medically under-resourced developing countries, where more than 80 percent of the world’s population lives, breast FNAB is the test for all palpable lesions, in a setting where preoperative imaging and core needle biopsy and histopathology are not readily available and are expensive options.5,6
Challenges and solutions
The greatest challenge now to breast FNAB cytology is the quality of the FNAB procedure and of the smearing technique. These are crucial to a successful breast cytology service and the major source of quality assurance problems with breast FNAB. Poor performance of the FNAB and the direct smear are the elephant in the room in any discussion of the role of breast cytology.2
Currently, radiologists and their trainees perform the majority of breast FNAB in the developed world, rather than cytopathologists. There is a long and successful tradition of cytopathologists carrying out FNAB of breast palpable lesions. Cytopathologists are immediately aware of the quality of their technique because they are reporting the slides, while radiologists often have minimal contact with the reporting pathologist. FNAB is regarded as a simple test, but it requires good training and ongoing experience with constant monitoring of the diagnostic yield and adequacy rates. The number of FNAB of breast has decreased in the developed world, resulting in fewer training opportunities for trainee radiologists and pathologists and a lack of adequate training. This has led to a general decrease in the quality of breast cytology specimens.
The radiologist uses ultrasound guidance for palpable and impalpable lesions, which has increased the range of accessible lesions but at the same time accentuated the problems in the performance of the FNAB and in making smears. The key elements in a breast FNAB are the fixation of the specimen and a rapid technique. Ideally, the needle should be introduced for fewer than 10 seconds, with 10 to 15 rapid passages of the needle into and just through the lesion, using the cutting action of the needle bevel. For cytopathologists and radiologists, ultrasound can be helpful in assessing palpable lesions. But it is more difficult to fix a palpable breast lesion when an ultrasound probe is present, and the time the needle is in the lesion, the “dwell time,” is lengthened, leading to an increased incidence of blood contamination and clotting of the material in the needle. Further, if aspiration is applied early in the FNAB without the cutting action of the needle having been utilized, the result may be inadequate, hemodiluted, and obscured material.
The second crucial preanalytical step is preparing direct smears, and poorly trained operators or their assistants can ruin good material by using poor smear-making technique. Liquid-based preparations do avoid poor smearing technique and the air drying of alcohol-fixed material, but they prevent ROSE, decrease the crucial pattern recognition diagnostic features in breast cytology, and increase expense.
ROSE carried out by a cytopathologist or well-trained cytotechnologist attending the FNAB procedure lowers inadequacy rates, makes it possible to provide an immediate provisional report to the clinician and patient, improves the quality of direct smears and the triaging of material for expensive ancillary tests, and decreases the costs of patient recalls and second procedures. Most important, ROSE provides immediate contact between the cytopathology team and radiologist. There can be constant interaction between the cytopathologist viewing the slides and the proceduralist, which leads to better quality FNAB material and reporting and better breast care for the patient. Ideally, the cytopathologist can perform the FNAB using ultrasound, if necessary, for palpable lesions or to target an impalpable lesion found by imaging. If this is not possible, a radiologist or other clinician willing to develop his or her technique and work with the cytopathologist to achieve better results should perform the FNAB.7
There are analytical challenges in interpreting breast cytology slides, and for inexperienced pathologists this is particularly so. Breast FNAB cytology requires specific training and continuing exposure to a significant caseload, just as in any other specialty area of cytology or surgical pathology. The reduction in the number of cases in most teaching hospital programs has led to a reduction in the level of training in breast cytopathology. In breast cytology, high cellularity and dispersal do not necessarily mean malignancy, and proliferative lesions and intraductal and even invasive carcinomas can have overlapping features. Distinguishing intraductal and invasive carcinomas can be difficult.
Core needle biopsy (CNB) in some parts of the developed world has virtually replaced breast FNAB, particularly in mammographic screening program assessment clinics, where a significant proportion of the cases involve workup of calcifications, and in the follow-up of any abnormal mammogram in general breast work. Breast core reporting is part of most surgical pathology practices and the core biopsy technique is relatively standard, so no special training is required. Further, the screening program experience in the use of FNAB and CNB has been inappropriately extrapolated into the assessment of all breast lesions, whether palpable or not, in clinical breast units. There is a need to establish and recommend the most appropriate use of these two complementary tests. CNB is more invasive with a greater rate of complications and is not required in many cases in the general breast clinic where the vast majority of lesions will be cysts, fibrocystic change, fibroadenomas, and other mass lesions including a small number of carcinomas. Breast FNAB, particularly when ROSE is available, can be used to triage the cases that do require CNB, leading to reduced costs. CNB is more expensive in terms of the biopsy equipment and the histopathology processing and reporting, and as such is inappropriate in a low-resource setting.5,6
Standardized reporting system
It is in this environment of the changing role of breast FNAB cytology that the members of the IAC Breast Group are developing a standardized reporting system. Small groups have prepared draft reviews and statements on the technical aspects of the FNAB; the different diagnostic categories used in reporting cases, including definitions, suggested terminology, and risk of malignancy based on positive and negative predictive values; and on the appropriate current ancillary testing role. These drafts have just been distributed to all members of the group for discussion. A consensus will be reached and draft documents will then be published on the IAC website early this year. Cytopathologists and clinicians will have the opportunity to critique and discuss the drafts, and the group will address their comments and modify the drafts where appropriate. Discussions will then be held with other cytology organizations to achieve, if possible, an international consensus. The rationale for the reporting system was presented at a number of cytology meetings in 2016 and 2017. The final documents will be published this year, and an atlas will be published by end of the year.
The standardized approach will include best-practice guidelines for the FNAB and smear-making techniques and the structure of reports. Structured reporting improves the quality, clarity, and reproducibility of reports across departments and between states and countries, and it will improve patient management and facilitate research and quality assurance measures.8 Standardized use of cell blocks, IHC, ISH, and other molecular tests of prognostic and diagnostic markers will improve care and reduce costs.
Structured reports establish a format with standard headings, definitions, and common terminology and include required information, which can be either a mandatory standard or a recommended guideline.9 They are usually based on a checklist that matches the workflow of the laboratory and cytopathologist and are presented in a clear format that conveys information across borders to pathologists and to clinicians.
The FNAB cytology report should resemble a breast core or any other surgical report and include: minimum data requirements, which can include a statement of whether the lesion is completely benign; a statement of cellularity, which is a measure of the adequacy of the material; a cytological description, which should include key cytological criteria; and a conclusion or summary using a standardized descriptive terminology diagnosis. This conclusion should be as specific as possible or, if a specific diagnosis is not possible, provide a weighted differential diagnosis based on the cytological criteria present. A code or category can be part of the body of the report and is useful for quality assurance and research, but a simple number should never be used in isolation or as a conclusion, as it will impair the clinician’s understanding of the individual report. The key to the report is a clear, descriptive diagnosis using standardized terminology.
The Breast Group has decided to use a five-category system used widely internationally: category 1: insufficient material; category 2: benign; category 3: atypical, probably benign; category 4: suspicious for malignancy, probably in situ or invasive carcinoma; and category 5: malignant.
There has been discussion around the use of the terms “insufficient” or “inadequate” for cases that lack epithelium, such as cyst contents, and around the definitions of “atypical” and “suspicious for malignancy” and the various situations when these terms should be used. The decision was made to retain an “atypical” category, which allows for a high NPV for a benign diagnosis, and a “suspicious for malignancy” category, to maintain a high PPV for a malignant diagnosis—and these two categories allow for stratification of the risk of malignancy. The causes of an “atypical” cytological diagnosis include technical problems with the FNAB and the smear making, scant material and interpretive problems related to the inherent characteristics of the lesion, or a combination of these factors intertwined with the experience of the cytopathologist. The causes of a “suspicious for malignancy” diagnosis are similar and should always be stated in the report along with the specific lesion the smears are suspicious of.
A structured reporting system requires checklists of key cytological features for specific lesions that are based on an analytical approach using low-power pattern recognition combined with high-power cytological features integrated in a final diagnosis.10
The FNAB cytology report is used in conjunction with the clinical and imaging findings in the triple-test approach, which yields very high PPV and NPV and provides the basis for management decisions. The Breast Group will establish best-practice protocols for the suggested management of each of the five categories with their varying risks of malignancy, while taking into account the vast differences between the developed and developing world in the potential availability of imaging, CNB, surgical pathology, and management options. These best-practice guidelines will include the indications for and role of FNAB and CNB in the management algorithms and allow for the great variations in medical infrastructure.5,6
For example, the current draft document suggests an “atypical” report should lead to an immediate reassessment of the imaging and clinical findings. If the triple test is negative apart from the atypical cytology report, a decision can be made to simply review the patient at a shortened time interval. Or if the imaging or clinical findings are indeterminate, immediate CNB can be performed. Where CNB is not available, repeat FNAB or possibly excision biopsy can be the management option. On the other hand, a “suspicious” cytology report requires a mandatory biopsy, which can be a repeat FNAB but is usually a CNB if available, or in some situations a simple excision biopsy.
Members of the International Academy of Cytology Standardized Reporting of Breast FNAB Cytology Group hope that cytopathologists and cytotechnologists will review the draft proposals and provide their input once the proposals are placed on the IAC website.
- Field AS, Schmitt F, Vielh P. IAC standardized reporting of breast fine-needle aspiration biopsy cytology. Acta Cytol. 2017;61(1):3–6.
- Field AS. Breast FNA biopsy cytology: current problems and the International Academy of Cytology Yokohama standardized reporting system. Cancer Cytopath. 2017;125(4):229–230.
- Abati A, Abele J, et al. The uniform approach to breast fine-needle aspiration biopsy. Diagn Cytopathol.1997;16(4):295–311.
- Schmitt F, Vielh P. Fine-needle aspiration cytology samples: a good source of material for evaluating biomarkers in breast cancer. Histopathology. 2015;66(2):314–315.
- Masood S, Vass L, Ibarra JA Jr, et al. Breast pathology guideline implementation in low- and middle-income countries. Cancer. 2008;113(8 suppl):2297–2304.
- Anderson BO. Fine-needle aspiration for breast cancer diagnosis: one size does not fit all. J Natl Compr Canc Netw. 2016;14(5):599–600.
- Ljung BM, Drejet A, Chiampi N, et al. Diagnostic accuracy of fine-needle aspiration biopsy is determined by physician training in sampling technique. Cancer. 2001;93(4):263–268.
- Ellis DW, Srigley J. Does standardised structured reporting contribute to quality in diagnostic pathology? The importance of evidence-based datasets. Virchows Arch. 2016;468(1):51–59.
- Structured Pathology Reporting of Cancer. Royal College of Pathologists of Australasia website. https://www.rcpa.edu.au/Health-Care-Professionals/Structured-Pathology-Reporting-of-Cancer.
- Field AS, Zarka MA. Chapter 5: Fine needle aspiration biopsy cytology of breast: a diagnostic approach based on pattern recognition. In: Practical Cytopathology: A Diagnostic Approach to Fine Needle Aspiration Biopsy.Philadelphia: Elsevier; 2017.
For me, friendships formed and lessons learned have more than compensated for the effort invested over the years on CAP committees, but make no mistake: When we meet, what we’re doing is work. The professional engagement is enjoyable, but a person can get tired toward the end of a two-day meeting, not to mention homework in the evenings.
At such times, I can always reboot by resurrecting the memory of a tumor board meeting about six months after we introduced the first CAP cancer protocols in our practice. Feedback had been sparse; we’d had no complaints, and that was good. But I knew what had gone into those protocols. At the very least, I felt, we had earned a banner over the hospital entrance.
Well, as we say in my family, if you don’t know, ask. So one day at tumor conference, I asked the crew of oncologists, internists, radiation oncologists, radiologists, oncologic surgeons, general surgeons, breast surgeons, gynecologic oncologic surgeons, urologists, family practitioners, and others sitting in the room what they thought about the new protocols. My question elicited an overwhelmingly positive response—the most positive response I’ve ever gotten in tumor conference to anything I’ve ever said! Everybody agreed the reporting templates had greatly increased the effectiveness of communication between pathologists and physicians in all these fields because they could now quickly find what they were looking for. We had 13 different pathologists writing narrative reports back then, and it seems the clinicians often had trouble finding what they needed. Now everyone knew where to find their nugget and could be certain it was included. That was a good tumor conference.
All of which is to say that I am a big believer in the cancer protocols. They are based, of course, on the American Joint Committee on Cancer staging manual, the most recent edition of which was implemented Jan. 1. We’ve had enough time with it now to confirm that the new edition is more ambitious, impressive, and perhaps even more groundbreaking than its predecessors.
The eighth edition features 12 entirely new staging systems and reflects the input of a greatly expanded team of international experts. It comprises 83 chapters, was three years in the making, and represents the work of 434 individuals from six continents, 23 countries, and 188 institutions. There is a searchable electronic version incorporating additional staging forms and other supplementary resources. It’s a big step up.
Anyone who has given the first chapter a thoughtful read (or watched the Dec. 14, 2017 CAP webinar given by Mahul B. Amin, MD, editor-in-chief of the staging manual, and Thomas P. Baker, MD, who chairs the CAP Cancer Committee) knows that this edition creates a new and inclusive gestalt. Members of 18 expert panels represent every cancer specialty. Another seven “cores” are home to experts in crosscutting disciplines such as precision medicine, evidence-based medicine and statistics, and content harmonization. These multidisciplinary resource teams bridge all disciplines, advising and encouraging transparent and useful give-and-take around staging, characterization, and utility.
I am pleased to note that Dr. Amin, professor and chair, Department of Pathology and Laboratory Medicine, and Gerwin endowed chair for cancer research, University of Tennessee Health Science Center, is a former chair of the CAP Cancer Committee. The decision to appoint Dr. Amin editor-in-chief says good things about the American College of Surgeons, which provides administrative support to the AJCC. Their choice of Dr. Amin, the first pathologist (in eight editions spanning almost five decades) to take on this honorable and pivotal task, is a credit to our specialty.
One reason, no doubt, is his ability to articulate the integral role of anatomic and clinical pathologists in cancer diagnosis and treatment. Dr. Amin knows how to explain to nonpathologists that our ability to work with the tools of molecular diagnostics enables more accurate subclassification at the patient care level. He can capture the ways that pathologists understand data mining and how it is employed to investigate potential treatment alternatives. He can help other specialists understand just what we do.
Whenever the AJCC releases a new edition of the staging manual, the CAP Cancer Committee uses it to create or revise the CAP cancer protocols. Subspecialty teams manage the 63 CAP protocols and 14 biomarker templates. We released last summer the revised versions of 52 CAP cancer protocols that harmonize with the eighth edition.
The cancer protocols are a big project within the CAP; they are among the best things we do for our patients and our specialty. The Cancer Committee reports to the CAP Council on Scientific Affairs, chaired by Raouf Nakhleh, MD. Volunteers on the CSA and the Cancer Committee support Dr. Baker and Dr. Nakhleh to ensure our protocols provide all necessary information without burdening pathologists with irrelevant reporting criteria. Because the CAP volunteers who write the protocols are practicing pathologists, they know that sometimes too much is too much. Succinct, synoptic reports enable our members to give managing clinicians a sharp, quick, complete picture of what they need to know to apply the new staging system at the point of care.
I am writing this on New Year’s Eve and thinking about all the ways the new edition offers a fresh look at our lives and our work. We can use the term “paradigm shift” only in retrospect, but I predict we will come to see the release of the eighth edition of the AJCC Cancer Staging Manual as a pivotal moment in the science of cancer care. This revision offers all of us—pathologists, surgeons, clinicians, and patients—the vocabulary and context required to communicate clearly and comfortably about cancer diagnosis, prognosis, and treatment.
A Vaccine for Recurrent Urinary Tract Infections in Women: The Future or a Flash in the Pan?
The report in BJU International by Yang and Foley suggests that we may finally be on the right track in developing a genitourinary mucosal vaccine to prevent recurrent urinary tract infections (UTIs) in women. A urology history lesson in regard to our understanding of UTIs is needed to understand this “new” approach to the prevention of UTI; specifically, simple cystitis in women. Observant physicians in the 19th century, Robert Bentley Todd and William Osler, described the clinical course of a woman who developed cystitis before the introduction of antibiotics. If a patient with simple cystitis did not develop urosepsis or pyelonephritis and survived, the physician induced trauma of bleeding, cupping, leeches, enemas, high-dose opioids, and quinine as well as ingestion of nitric, benzoic, and even sulfuric acid, or even oil of turpentine; she would arise from bed 3 to 4 weeks later, weak and tired, but well with almost no reported possibility of developing another UTI in her lifetime.
It was Guyon, the famous French physiologist and urologist who, understanding the microbiological etiology of cystitis, proposed that women who survived the bladder infection had a reduced risk of recurrent infection due to “acquired immunity … the result of autovaccination from the absorption of toxins or bacteria in a state of modified virulence.” With the introduction of antibiotics, urologists, even Campbell in his 1956 Textbook of Urology, believed that “UTIs will soon be relegated to the waste basket of medical history,” an idea that continued into the golden age of antibiotics, the 1960s, when it was still believed that “the time has come to close the book on infectious diseases. We have basically wiped out infection in the United States” (William Stuart, Surgeon General of the United States of America, 1967). Well, that did not happen, and, in fact, with the clinical strategy of early antibiotic therapy, we may have inadvertently caused the contemporary problem of recurrent UTI in women by not allowing women with cystitis to develop their own preventive immunity. Antibiotics, even prophylatic antibiotics, cannot solve this problem and, in fact, may be contributing to future problems with development of antibiotic-resistant uropathogens and dramatic changes in our human microbial biodiversity.
So, will vaccines be the key? Certainly, the report by Yang and Foley is encouraging; however, I will go out on a limb and predict that, while we will see significant benefits with improved vaccine development, these benefits will be temporary for individuals (and perhaps even for populations) as the individual (and population-wide) microbiome adapts to the new but rather crude vaccine. It will only be further understanding of the relationship between our changing microbiome and our genome that will lead to ultimate prevention of UTIs, and that strategy might very well include a vaccine approach. In the interim, I personally will be very excited to use this vaccine in my patients and hope that the results of the ongoing international randomized placebo-controlled study will be positive.
The presence of a cardiovascular implantable electronic device has long been a contraindication for the performance of magnetic resonance imaging (MRI). We established a prospective registry to determine the risks associated with MRI at a magnetic field strength of 1.5 tesla for patients who had a pacemaker or implantable cardioverter–defibrillator (ICD) that was “non–MRI-conditional” (i.e., not approved by the Food and Drug Administration for MRI scanning).
Patients in the registry were referred for clinically indicated nonthoracic MRI at a field strength of 1.5 tesla. Devices were interrogated before and after MRI with the use of a standardized protocol and were appropriately reprogrammed before the scanning. The primary end points were death, generator or lead failure, induced arrhythmia, loss of capture, or electrical reset during the scanning. The secondary end points were changes in device settings.
MRI was performed in 1000 cases in which patients had a pacemaker and in 500 cases in which patients had an ICD. No deaths, lead failures, losses of capture, or ventricular arrhythmias occurred during MRI. One ICD generator could not be interrogated after MRI and required immediate replacement; the device had not been appropriately programmed per protocol before the MRI. We observed six cases of self-terminating atrial fibrillation or flutter and six cases of partial electrical reset. Changes in lead impedance, pacing threshold, battery voltage, and P-wave and R-wave amplitude exceeded prespecified thresholds in a small number of cases. Repeat MRI was not associated with an increase in adverse events.
In this study, device or lead failure did not occur in any patient with a non–MRI-conditional pacemaker or ICD who underwent clinically indicated nonthoracic MRI at 1.5 tesla, was appropriately screened, and had the device reprogrammed in accordance with the prespecified protocol. (Funded by St. Jude Medical and others; MagnaSafe ClinicalTrials.gov number, NCT00907361.)
The use of magnetic resonance imaging (MRI) poses potential safety concerns for patients with an implanted cardiac device (cardiac pacemaker or implantable cardioverter–defibrillator [ICD]). These concerns are a consequence of the potential for magnetic field–induced cardiac lead heating, which could result in myocardial thermal injury and detrimental changes in pacing properties.1-3 As a result, it has long been recommended that patients with an implanted cardiac device not undergo MRI scanning, even when it otherwise may be considered to be the most appropriate diagnostic imaging method for the patient’s clinical care.4
Over the past two decades, cardiac devices have been designed to reduce the potential risks associated with MRI.5,6 Such devices, if they have been shown to pose no known hazard under certain specified conditions, are labeled “MRI-conditional” by the Food and Drug Administration (FDA) Center for Devices and Radiological Health. However, it is estimated that 2 million people in the United States and an additional 6 million worldwide7 have devices that have not been shown to meet these criteria and are therefore considered “non–MRI-conditional.” At least half the patients with such devices are predicted to have a clinical indication for MRI during their lifetime after device implantation.8
The MagnaSafe Registry was established to determine the frequency of cardiac device–related clinical events and device setting changes among patients with non–MRI-conditional devices who undergo nonthoracic MRI at a magnetic field strength of 1.5 tesla, as well as to define a simplified protocol for screening, monitoring, and device programming for such patients.
The MagnaSafe Registry was a prospective, multicenter study involving patients with a non–MRI-conditional pacemaker or ICD who underwent a clinically indicated, nonthoracic MRI examination at 1.5 tesla. The rationale, design, and protocol have been described previously.9 The protocol, which is available with the full text of this article at NEJM.org, was written after consultation with personnel at the Center for Devices and Radiological Health of the FDA, who requested that thoracic scans be excluded because of a higher perceived risk of adverse outcomes. An investigational device exemption was obtained in April 2009 for the purpose of data collection. All participating centers obtained approval from a local or independent institutional review board.
None of the funders of the study had any role in the design of the study protocol, in the collection or analysis of the data, or in the writing of the manuscript. The authors had full access to the data, performed the analyses, and vouch for the completeness and accuracy of the data and for the fidelity of the study to the protocol.
Patients were included in the registry if they were 18 years of age or older and had a non–MRI-conditional pacemaker or ICD generator, from any manufacturer, that was implanted after 2001,10 with leads from any manufacturer (without implantation date limitation), and if the patient’s physician determined that nonthoracic MRI at 1.5 tesla was clinically indicated (see Tables S1 and S2 in the Supplementary Appendix, available at NEJM.org, for a list of pacemaker and ICD manufacturers and models). The exclusion criteria were an abandoned or inactive lead that could not be interrogated, an implanted device other than a pacemaker or an ICD, an MRI-conditional pacemaker, a device implanted in a nonthoracic location, or a device with a battery that was near the end of its battery life (with a device interrogation display that read “elective replacement indicator”). In addition, pacing-dependent patients with an ICD were excluded because it was not possible to independently program tachycardia and bradycardia therapies for all ICD models at the time of study design. All participants provided written informed consent.
During the first 2 years of the study, the Centers for Medicare and Medicaid Services National Coverage Determination (NCD) stated that a patient with a pacemaker or an ICD was not eligible for coverage for MRI. In March 2011, a change to the NCD was granted that allowed reimbursement for patients enrolled in a prospective registry designed to determine the risk associated with MRI.11
MRI PROTOCOL AND MONITORING
All studies were performed in a 1.5-tesla MRI scanner; there was no vendor restriction (a list of manufacturers and models is included in Table S3 in the Supplementary Appendix). A physician, nurse practitioner, or physician assistant with cardiac device expertise and training in advanced cardiac life support was in attendance. Blood pressure, pulse oximetry, and cardiac rhythm were monitored with an MRI-compatible system from the time of device reprogramming until restoration of baseline values. Further details are provided in the MagnaSafe Protocol section of the Supplementary Appendix.
DEVICE INTERROGATION AND PROGRAMMING
Prescanning device interrogation was performed with the use of a standardized protocol (Figure 1).9 If the patient was asymptomatic and had an intrinsic heart rate of at least 40 beats per minute, the device was programmed to a no-pacing mode (ODO or OVO). Symptomatic patients or those with an intrinsic heart rate of less than 40 beats per minute were determined to be pacing-dependent, and the device was reprogrammed to an asynchronous pacing mode (DOO or VOO). For non–pacing-dependent patients with an ICD, all bradycardia and tachycardia therapies were inactivated before the MRI. Pacing-dependent patients with an ICD were excluded, because not all ICD models allowed for independent inactivation of tachycardia and bradycardia therapies. After the MRI, baseline settings were restored, full device interrogation was repeated, and if necessary, the device was reprogrammed to maintain adequate pacing and sensing. Further details are provided in the MagnaSafe Protocol section of the Supplementary Appendix.
PRIMARY AND SECONDARY END POINTS
The primary end points, which were assessed during or immediately after the MRI examination, were death, generator or lead failure requiring immediate replacement, loss of capture (for pacing-dependent patients with pacemakers), new-onset arrhythmia, and partial or full generator electrical reset. The secondary end points, which were assessed immediately after the MRI examination and at the final follow-up, were a battery voltage decrease of 0.04 V or more, a pacing lead threshold increase of 0.5 V or more,12 a P-wave amplitude decrease of 50% or more, an R-wave amplitude decrease of 25% or more and of 50% or more,13 a pacing lead impedance change of 50 ohms or more,14 and a high-voltage (shock) lead impedance change of 3 ohms or more.
Patients with any secondary end-point event were required to undergo repeat device interrogation within 7 days, at 3 months (±30 days), and at 6 months (±30 days) after the MRI to determine whether the device settings had returned to baseline. If a secondary end-point event did not occur, a single device interrogation was required at between 3 and 6 months after the MRI (±30 days). Patients who had a primary end-point event were seen in follow-up at the discretion of the supervising physician. Further details and definitions of end points are provided in the Supplementary Appendix.
A case was defined as an instance in which a patient who provided informed consent entered the scanner and underwent MRI of one or more anatomical regions during a single examination session. If the patient returned on a subsequent day for repeat MRI, a separate informed consent was obtained and the data were entered as a unique case.
The mean (±SD) yearly rate of device replacement due to spontaneous malfunction has been estimated to be 0.46±0.22% for pacemakers and 2.07±1.16% for ICDs.15 Using these estimates and assuming a device failure rate during or after MRI of 0, we determined that 1000 cases in which patients had a pacemaker (pacemaker cases) and 500 cases in which patients had an ICD (ICD cases) would be needed to yield a 95% confidence interval of 0 to 0.5% for pacemakers and 0 to 1.0% for ICDs.
Data were analyzed separately for the pacemaker and ICD cohorts with the use of R statistical software, version 220.127.116.11 The decision not to perform statistical comparisons between the pacemaker and ICD cohorts was made before enrollment began. The Wilson score method without continuity correction was used to calculate 95% confidence intervals for single proportions for primary end-point events. The linear association between lead age and each of the secondary end points was assessed with Pearson’s product moment correlation coefficient.
STUDY PATIENTS AND FOLLOW-UP
From April 2009 through April 2014 at 19 centers in the United States, clinically indicated nonthoracic MRI was performed in a total of 1000 pacemaker cases (818 patients) and 500 ICD cases (428 patients). The baseline characteristics of the patients are shown in Table 1. Follow-up data, which included data from a full device interrogation, were available in 1395 cases (93%) at 6 months. Additional information about the study population is provided in the Supplementary Appendix.
MRI PROCEDURAL DATA
A total of 75% of the MRI examinations were performed on the brain or the spine. The mean time patients spent within the magnetic field was 44 minutes. During the MRI examination, four patients reported symptoms of generator-site discomfort; one patient with an ICD was removed from the scanner when a sensation of heating was described at the site of the generator implant, and the patient did not complete the examination. No patient with generator-site symptoms had the device placed within the “field of view” (the MRI imaging area), had a study end-point event, or reached the specific absorption rate limit set by the FDA for the scanned body site.
PRIMARY END POINTS
There were no deaths, lead failures requiring immediate replacement, or losses of capture during the MRI examination among patients who were appropriately screened and had their device reprogrammed for imaging (Table 2). In one patient with an ICD who was not pacing-dependent, antitachycardia therapy was left in the active mode during the MRI (a protocol violation). During the post-MRI evaluation, the ICD could not be interrogated, and immediate generator replacement was required. Further details are provided in the Supplementary Appendix.
Four patients had atrial fibrillation and two patients had atrial flutter during or immediately after the MRI (Table S4 in the Supplementary Appendix). Five of these patients had a history of paroxysmal atrial fibrillation and were receiving warfarin; two were receiving antiarrhythmic therapy. Three of the patients returned to sinus rhythm before leaving the MRI environment, and the remaining three patients returned to sinus rhythm within 49 hours. No ventricular arrhythmias were noted.
In six cases (five patients), the patient had partial generator electrical reset; in all six cases, the patients had pacemakers that had been implanted 5.7 to 9.7 years before the MRI (Table S5 in the Supplementary Appendix). Settings in the device memory that were reset included patient and device or lead identification information. No appropriately screened and reprogrammed device underwent a full electrical reset.
SECONDARY END POINTS
The results with regard to the secondary end points and measured differences between post-MRI and pre-MRI device settings for both pacemakers and ICDs are shown in Table 3 and as a histogram in Fig. S1 in the Supplementary Appendix. A decrease of 50% or more in P-wave amplitude was detected in 0.9% of pacemaker leads and in 0.3% of ICD leads; a decrease of 25% or more in R-wave amplitude was detected in 3.9% of pacemaker leads and in 1.6% of ICD leads, and a decrease of 50% or more in R-wave amplitude was detected in no pacemaker leads and in 0.2% of ICD leads. An increase in pacing lead threshold of 0.5 V or more was detected in 0.7% of pacemaker leads and in 0.8% of ICD leads.
A pacing lead impedance change of 50 ohms or more was noted in 3.3% of pacemakers and in 4.2% of ICDs. For both pacemakers and ICDs, any decrease in pacing lead impedance from baseline occurred in 54% of atrial leads and in 55% of ventricular leads, and any increase occurred in 19% of atrial and 22% of ventricular leads. However, when the change in pacing lead impedance was compared as a continuous variable with the change in P-wave or R-wave voltage or pacing lead threshold, no clinically significant correlations were noted (Table S6 in the Supplementary Appendix).
LEAD AND DEVICE AGE AND CLINICAL END POINTS
Among patients who had undergone placement of a new generator or lead within 90 days before the MRI, there were no primary end-point events, and secondary end-point events were limited to a change in pacing lead impedance in 2 of 53 new pacemaker leads and in 1 of 27 new ICD leads. Among patients with leads that had been placed more than 10 years before MRI, there were no primary end-point events, and secondary end-point events were noted in 1 of 31 ICD leads (impedance change of ≥50 ohms) and in 14 of 172 pacemaker leads (1 with a P-wave amplitude decrease of ≥50%, 1 with a pacing threshold increase of ≥0.5 V, and 11 with an impedance change of ≥50 ohms). When the continuous variables of pacing lead threshold change, P-wave amplitude change, R-wave amplitude change, and impedance change were compared separately with the time since lead placement, no clinically significant correlations were found (Table S7 in the Supplementary Appendix).
PATIENTS WITH REPEAT MRI EXAMINATIONS
The maximum number of MRI examinations performed in patients in the MagnaSafe Registry was 11 in one patient with a pacemaker and 7 in one patient with an ICD (Table S8 in the Supplementary Appendix). The median interval between MRIs among patients who underwent more than one MRI examination was 153 days in patients with a pacemaker (range, 3 to 1309 days) and 91 days in patients with an ICD (range, 1 to 1376 days). In the examination of secondary end points, we found no clinically important differences between cases in which the patient underwent a single MRI and cases in which patients had undergone a previous MRI (Table S9 in the Supplementary Appendix).
PERSISTENT CHANGES IN DEVICE SETTINGS
Patients whose cardiac device exceeded the limit for a change in setting at the time of the MRI (a secondary end-point event) were asked to return for a repeat interrogation within 7 days and at 3 months and 6 months (pacemakers, 11% of cases; ICDs, 26% of cases). The proportions of cases in which there were persistent changes in device settings at the final follow-up are shown in Table 4. A higher incidence of long-term setting changes was seen with ICDs than with pacemakers. A long-term battery voltage decrease of 0.04 V or more occurred in 4.2% of ICD cases, and a long-term high-voltage lead impedance change of 3 ohms or more occurred in 10.0% of ICD cases.
In this study, we investigated the use of nonthoracic MRI at 1.5 tesla in patients with an implanted non–MRI-conditional cardiac device (pacemaker or ICD). We implemented a specific protocol for device interrogation, device programming, patient monitoring, and follow-up that was designed to reduce the risk of patient harm from MRI effects. In our study, no patient who was appropriately screened and had the device reprogrammed in accordance with our protocol had a device or lead failure. In one case, an ICD that was not properly reprogrammed before the MRI could not be interrogated after the procedure, and immediate generator replacement was required. In six cases, atrial arrhythmias occurred, each lasting less than 49 hours; six partial electrical resets occurred that were detected and corrected during post-MRI reprogramming. Changes in device settings were common, but relatively few exceeded our prespecified threshold criteria for a clinically important change; the most common change was a 3-ohm change in ICD high-voltage (shock) lead impedance (16.4% of cases).
When pre-MRI and post-MRI battery voltage measurements were compared, a small decrease was noted for both pacemakers and ICDs. The radiofrequency energy generated during MRI scanning creates a temporary decrease in battery voltage, which has typically been reported to resolve after several weeks. In our study, all pacemaker voltage decreases of 0.04 V or more had resolved at the last follow-up, although some ICD voltage decreases of 0.04 V or more had not.
At the time that the study was being designed, we did not anticipate the demand for repeat MRI for patients with an implanted cardiac device. If exposure to a strong radiofrequency field resulted in substantial thermal injury at the lead–myocardial interface,1 these patients should be at the greatest risk for a cumulative detrimental change in pacing properties. The only indication of such an effect in our study was a higher rate of high-voltage (shock) lead impedance changes among patients who had had previous MRI than among those who had not had previous MRI (21.5% vs. 14.9%).
Several smaller studies examining the risk associated with MRI in patients with an implanted device have reported varying effects on cardiac device settings.17-31 On the basis of this early experience, position statements recommended caution in the performance of MRI in patients with an implanted cardiac device.32,33 Subsequently, a larger prospective study examined 555 cases of scanning (including thoracic imaging) to assess the risk associated with MRI; no adverse clinical events occurred among the patients who underwent MRI, and the observed setting changes did not require device revision or reprogramming.7
Although it has been suggested that implanted generators and leads may be removed and then replaced to allow for MRI, such procedures may have greater risks than those associated with nonthoracic MRI in the current study. The rate of major complications among patients undergoing generator replacement with or without the placement of an additional transvenous lead was 4 to 15% in a prospective registry.34 In addition, single-center and multicenter studies have shown a rate of major complications associated with elective laser-assisted lead extraction that is in the range of 0.4 to 2%.35-38 Thus, device removal and replacement seem unlikely to be safer than proceeding with scanning for patients with a pacemaker or an ICD who require a nonthoracic MRI, provided a protocol similar to the one used in our study is followed.
The limitations of this study should be considered carefully. This registry represents a heterogeneous experience, with generators and leads from multiple manufacturers and initial as well as repeat examinations at 1.5 tesla. Thus, the results may not be predictive of findings with all device–lead combinations or higher MRI field strengths. Also, because patients younger than 18 years of age and MRI examinations of the thorax were excluded and the number of left ventricular leads was relatively small, it may not be possible to extrapolate the current data to a pediatric population, to patients undergoing MRI of the chest, or to patients with cardiac resynchronization devices. Finally, we excluded pacing-dependent patients with an ICD, because not all such patients had a device that was capable of providing pacing function while allowing for inactivation of tachycardia therapy. Therefore, our method should not be applied to pacing-dependent patients with an ICD unless independent programming of the bradycardia and tachycardia functions is possible.
In conclusion, we investigated the use of nonthoracic MRI at 1.5 tesla in patients with an implanted non–MRI-conditional cardiac device. No patient who was appropriately screened and had the cardiac device reprogrammed according to our protocol had device or lead failure. Substantial changes in device settings were infrequent and did not result in clinical adverse events.
Presented in part as a Late Breaking Clinical Trial at the American Heart Association Annual Scientific Sessions, Chicago, November 15–19, 2014.
Supported by grants from St. Jude Medical, Biotronik, Boston Scientific, and the Hewitt Foundation for Medical Research, and by philanthropic gifts from Mr. and Mrs. Richard H. Deihl, Evelyn F. and Louis S. Grubb, Roscoe E. Hazard, Jr., and the Shultz Steel Company.
Disclosure forms provided by the authors are available with the full text of this article at NEJM.org.
Lungs and pleural spaces are clear. Cardiomediastinal contours are normal. The bones are diffusely sclerotic and there are H-shaped vertebral bodies in keeping with sickle cell anaemia.
With the help of artificial intelligence, NASA’s Frontier Development Lab and Intel are mapping the moon’s craters to find hidden lunar resources.
Scientists believe the moon is rife with natural resources that could help space explorers settle the lunar landscape – much like early settlers did on earth.
But before they can access those resources, they need to find them.
“We have 50 years’ worth of NASA imagery from all sides of the moon,” said Shashi Jain, innovation manager at Intel’s Software and Services Group. “We’ve only recently begun to combine them and make one big, awesome map.”
Working with the NASA Frontier Development Lab (FDL), a team of Intel AI engineers and data scientists are tackling the challenge of building complex maps of the lunar poles.
Craters in the permanently shadowed polar regions of the moon are potentially filled with water, ice and other volatile resources that can be used to produce rocket fuel, an air supply for astronauts or other essential materials, according to Jain.
The shadowy lunar surface creates artifacts in NASA images, which make it difficult to accurately map potential landing sites for lunar prospectors.
Crews on long exploratory missions to outer space can’t carry all the resources they need, so finding things like water, hydrogen, carbon dioxide, nitrogen and methane may help NASA plan future missions to the moon or even to Mars.
Making Maps from Millions of Images
It turns out that making maps from lunar data is hard, said Jain. Planetary scientists get strips of imagery from orbiting satellites, which are at different lighting angles, scales and types. They have to manually line them up using craters and other features as landmarks. If any strips are out of alignment or too dark, the result is a poor quality map.
Deep learning, a branch of machine learning that uses neural network models to understand large amounts of data, could speed up the process of mapping the moon.
Whereas machine learning allows machines to act or think without being explicitly directed to perform specific functions, deep learning can accelerate processes like image recognition, quickly identifying and mapping craters and other obstacles on the moon.
“Space data is often massive, multidimensional and dynamic,” said James Parr, director of NASA FDL.
It’s critical for scientists to quickly process ever-evolving lunar data to help guide plans for future missions, according to Parr.
To get started, the team first needed to create a computer vision algorithm and train it to identify craters. NASA FDL and Intel built a crater image training set using 30,000 images. It took Jain six hours to manually find images containing a crater — but fully mapping the moon means looking at hundreds of millions of images.
In order to create detailed lunar maps, the team used two datasets from the NASA Lunar Reconnaissance Orbiter (LRO) mission — one set with optical images and the other with elevation measure data. Overlaying the two datasets created highly accurate maps, said Jain.
Since lunar craters acted as critical registration points to align the two datasets into one unified map, the team developed a computer vision algorithm to quickly and reliably identify craters.
The team automated lunar crater detection with 98.4 percent accuracy. By running their algorithm on the Intel Nervana Cloud, it took only one minute to classify 1,000 images, which is 100 times faster than human experts. The algorithm is also available in GitHub for use by other research teams.
Partners in Space Research
The NASA FDL space resource project was completed during a whirlwind eight-week program at the SETI Institute in Mountain View, California.
The space resource team was just one of five teams that took part in challenges in the summer program. Other teams tackled planetary defense and space weather challenges, like long period comets, radar 3D shape modeling, solar-terrestrial interactions and solar storm prediction.
“It was the summer of exploration with artificial intelligence right here on Earth,” said Jain.
With AI-generated maps of the moon’s poles, soon NASA will have summers, winters and years of exploration on the moon and beyond.