Respiratory Tract Diseases That May Be Mistaken for Infection
- 389 Downloads
The differential diagnosis of lung infiltrates in transplant patients includes non-infectious processes that may mimic pneumonia. Hydrostatic and non-cardiogenic pulmonary edema, as well as transfusion-related lung injury, may cause bilateral airspace opacification that may be confused with an infectious process. Chemotherapeutic agents, whether administered for treatment of an underlying hematologic malignancy, for induction prior to transplant, or for the treatment of GVHD or graft rejection after transplant, may cause lung injury that in some cases may not become apparent until years later. Radiation-induced lung injury may cause a distinct pattern of lung injury, which may present months after exposure. In hematopoietic stem cell transplant patients, pulmonary diagnoses such as engraftment syndrome, idiopathic pneumonia syndrome, and diffuse alveolar hemorrhage can mimic pneumonia. After lung transplantation, infiltrates may be seen as a consequence of primary graft dysfunction or allograft rejection. Organizing pneumonia may be seen both in recipients of solid organ and hematopoietic stem cell transplants. Organizing pneumonia may be cryptogenic or associated with the use of the mTOR inhibitor sirolimus. Failure to accurately diagnose non-infectious infiltrates may lead to unnecessary antibiotic treatment and failure to address the underlying pathophysiologic process.
KeywordsPneumonia Pneumonitis Infiltrate Radiographic Radiologic
The prevention, diagnosis, and treatment of pneumonia are critical to outcomes among transplantation patients. The diagnosis of pneumonia is usually considered when a new radiographic infiltrate is identified. Rarely, pneumonia may present initially without infiltrates in an immunocompromised host. However, even in severely immunosuppressed patients, infectious microbes cause inflammation through innate immune mechanisms that lead to tissue edema which is evident radiographically. The occurrence of pneumonia that is not apparent when a CT scan is included in the assessment is sufficiently rare that it will not be further considered here. Instead, we begin with an abnormal imaging study of the lungs, which is typically obtained during the workup of a symptom or sign such as cough, fever, or chills. The differential diagnosis of a lung infiltrate in a transplantation patient includes several common non-infectious causes. Failure to accurately diagnose non-infectious causes of lung infiltrates can lead to unnecessary treatment with antibiotics, and more importantly to failure to address the underlying pathophysiologic process. This chapter is focused on clinical presentations of pulmonary disorders that mimic infectious pneumonia.
Hematopoietic Stem Cell Transplantation (HSCT)
Despite advances in treatment regimens and supportive care, pulmonary complications still occur in up to 60% of HSCT recipients . These complications are mostly due to toxicities from conditioning regimens, delayed bone marrow recovery, prolonged immunosuppressive therapy, and graft-versus-host disease (GVHD). As the incidence of infectious pulmonary complications has diminished, largely due to effective prophylactic therapy, non-infectious pulmonary complications have emerged as a major cause of morbidity and mortality . Pulmonary complications have been divided into those that occur “early” (during the first 100 days after transplantation) and those that occur “late,” but this is not a rigid division. In particular, some “late” complications such as cryptogenic organizing pneumonia and constrictive bronchiolitis occur with substantial frequency during the first 100 days.
Non-cardiogenic pulmonary edema due to increased permeability of the alveolocapillary membrane (acute lung injury and adult respiratory distress syndrome (ARDS)) can occur as a result of a wide variety of causes in transplant patients. Sepsis is the most common cause of permeability edema of the lungs in general [6, 9, 10] and can cause radiographic infiltrates in transplant patients. In addition, transplant patients are susceptible to lung injury from causes unique to this population, such as from chemotherapy or the effects of acute GVHD. Transplant patients are also frequently exposed to treatments associated with lung injury in the general hospital population, such as transfusion. Transfusion-related acute lung injury (TRALI) is the leading cause of mortality from transfusions  and has been associated with all plasma-containing blood products, including immunoglobulins. The incidence of TRALI is not known, but has been estimated at 0.02% per unit transfused and 0.16% per patient transfused . Patients with TRALI commonly present with dyspnea, cough, fever, acute hypoxemia hypotension, and bilateral pulmonary infiltrates within 1–6 h after the transfusion. Transient leukopenia, due to pulmonary sequestration of the circulating pool of leukocytes, may be observed. The mainstay of treatment for TRALI is to discontinue the transfusion, followed by supportive care. Although there has never been a randomized controlled trial of glucocorticoid therapy, they have no effect on the 5–8% mortality. With supportive treatment, infiltrates usually resolve, within 96 h, and survivors have no long-term sequelae .
Engraftment Syndrome (ES)
ES is characterized by a constellation of symptoms and signs including fever, erythrodermatous skin rash, diarrhea, and non-cardiogenic pulmonary edema with bilateral pulmonary infiltrates, which generally occur within 5 days of neutrophil engraftment following HSCT. In more severe cases, systemic involvement, i.e., renal failure, hepatic failure, encephalopathy, or seizures, may be observed. Seen most often following autologous HSCT, ES has also been described in those individuals who have undergone allogeneic HSCT with a non-myeloablative preparative therapy. Although the pathophysiology of ES is not well understood, it is thought to result from a combination of endothelial injury due to preconditioning chemotherapy and the production and release of cytokines and products of neutrophil degranulation and oxidative metabolism, leading to capillary leak, with either local injury in the lung or systemic tissue injury . Bronchoalveolar lavage (BAL) may show a neutrophilic alveolitis. Surgical lung biopsies, when obtained, often reveal diffuse alveolar damage. Treatment entails observation and supportive care (i.e., antibiotics, intravenous fluids) in mild cases. High-dose corticosteroid therapy is very effective, often resulting in rapid clinical improvement in those with progressive or symptomatic ES. Respiratory failure requiring mechanical ventilation has been observed, however, in up to one-third of patients .
Idiopathic Pneumonia Syndrome (IPS)
In 1993, a panel convened by the NIH proposed a broad working definition of IPS as widespread non-lobar radiographic infiltrates in the absence of congestive heart failure or evidence of lower respiratory tract infection . IPS occurs in 10% of HSCT recipients, usually 14–90 days following transplantation. Mortality rates range from 50% to 70% . Possible etiologies of IPS include direct toxic effects of the chemoradiation conditioning regimen, occult infection, and/or the release of inflammatory cytokines secondary to some as yet unknown inciting stimuli. The association of IPS with the presence of acute GVHD after allogeneic HSCT suggests that alloreactive T cells may be at least one of these stimuli [17, 18].
The clinical presentation is non-specific, with symptoms of dyspnea, cough, and fever associated with diffuse infiltrates on chest radiograph. The diagnosis of IPS largely relies on the exclusion of infection on lower respiratory samples obtained from a diagnostic procedure, e.g., BAL or lung biopsy. Common pathologic findings of non-specific interstitial pneumonitis (NSIP) and/or diffuse alveolar damage (DAD) may be seen. Although no randomized controlled trials of treatment for IPS are available, current standards include high-dose intravenous corticosteroids and supportive care, such as supplemental oxygen and broad-spectrum antibiotics. Recent preclinical and clinical data suggest a potential role for tumor necrosis factor-α (TNF-α) in the pathogenesis of IPS [19, 20, 21], and a randomized trial using etanercept, a TNF receptor fusion protein, is being conducted by the Blood and Marrow Transplant Clinical Trials Network.
This same Network has included diffuse alveolar hemorrhage (DAH) within the definition of IPS, and we know of no reason to separate the two. Post-transplantation DAH was initially described in autologous HSCT recipients as widespread lung injury manifested by diffuse radiographic infiltrates that occurred in the absence of identifiable infection. DAH is now known to occur in both allogeneic and autologous transplant recipients and is seen in approximately 5% of all HSCT . The etiology is unclear, but is not clearly related to any specific coagulopathy or to thrombocytopenia . Pre-transplant high-dose chemotherapy, thoracic and/or total body irradiation, and undocumented infections are putative factors which may cause the initial injury, priming the lung for subsequent development of DAH. It can coincide with stem cell engraftment, but late onset (after the first 30 days) has been observed and is associated with a worse prognosis. Hemoptysis occurs in less than 20% of patients. Bronchoscopic diagnostic criteria include progressively bloodier returns on BAL or the presence of 20% or more hemosiderin-laden macrophages on cytologic inspection of BAL fluid. However, these bronchoscopic criteria may be seen in association with diffuse lung injury from a wide variety of causes, including infections, congestive heart failure, and malignancy. There are no prospective randomized trials addressing the treatment of DAH. Earlier retrospective studies demonstrated reduced need for mechanical ventilation and mortality in a cohort of patients receiving high-dose corticosteroids, but more recent observational studies found no survival benefit [24, 25].
Drug-Induced Lung Injury (DILI)
Chemotherapeutic agents characterized by class, indication, and pulmonary toxicity
HSCT patients receive high-dose chemotherapy as an induction regimen to eliminate marrow cells and prevent rejection of the graft. Following transplantation, immunosuppressive agents to prevent GVHD are prescribed for patients who received allogeneic grafts. Anti-thymocyte globulin is an antibody derived from rabbit or equine serum that can produce interstitial infiltrates and progress to ARDS [41, 42]. The mechanism of lung injury is not clear. As ATG is an anti-leukocyte antibody, the pathogenesis may be similar to that of transfusion-related lung injury (TRALI), or it may stem from its presence as a foreign protein [41, 42]. BCNU may produce DILI within 6 weeks of administration or as late as 20 years after administration to treat pediatric cancers [43, 44]. It presents with diffuse infiltrates and dyspnea, generally without fever, and is irregularly responsive to steroid therapy. The alkylating agents busulfan, melphalan, and cyclophosphamide can all produce a UIP/fibrosing pattern of injury . Busulfan was the first cytotoxic agent described to produce lung injury more than 50 years ago . In one series, the incidence of toxicity was 46% . Today, it is only used as an induction agent for HSCT. Melphalan is well-tolerated at standard doses but given at the high doses used for induction can produce a DIP-like presentation . Cyclophosphamide, used to treat lymphoma and for induction, can produce an early-onset pneumonitis, within 1–6 months of administration, which may be responsive to cessation of the drug or steroid therapy . It can also produce fibrosis and pleural thickening that can chronically progress despite cessation. Rituximab, which is a B-cell-depleting monoclonal antibody, is used to treat lymphomas and rheumatologic ailments. It can be used for induction as well. Symptoms may appear as early as 30 days or as late as 5 months. It rarely causes interstitial lung disease, with only 121 cases reported to date; however, 15% of the cases were fatal . Sirolimus, temsirolimus, and everolimus are mTOR inhibitors used for prevention of GVHD and for treatment of renal and other cancers [50, 51]. The mTOR inhibitors can all cause pneumonitis and are discussed in the subsequent section on solid organ transplantation.
Radiation-Induced Lung Injury
Total body irradiation (TBI) administered for induction is associated with acute pulmonary toxicity. Among 101 patients undergoing HSCT with TBI at Duke, one-third developed severe pulmonary toxicity, though the only independent factor correlated with the development of pulmonary toxicity was the number of chemotherapy regimens prior to transplant . Gopal et al. found a similar rate of severe pulmonary toxicity among patients receiving 12 cGy of TBI in 4 once-daily fractions (6 of 24 patients, 25%) . There was a lower incidence of severe toxicity among patients treated with 10.2 cGy in 6 twice-daily fractions (7 of 57, 12%); however, the difference was not significant (P-0.19). TBI has also been associated with alveolar hemorrhage in patients undergoing autologous transplantation .
Pulmonary Alveolar Proteinosis (PAP)
Cryptogenic Organizing Pneumonitis (COP)
Pathologically, COP is characterized by the presence of granulation tissue within the lumen of the distal air spaces with or without bronchoalveolar involvement (see Fig. 21.5b). This pathologic picture can be seen with multiple other accompanying diagnoses, such as congestive heart failure, infections, and drug-induced toxicity; hence, in the HSCT recipient, other diagnoses should be excluded before a diagnosis of COP is made. COP is highly responsive to corticosteroids. The minimal effective dose and duration of therapy are unknown; however, a prolonged steroid course with a slow taper is usually necessary due to high relapse rates. Macrolides have been used with success in some cases and might be considered in those individuals who are intolerant to steroid therapy or in whom relapse occurs . Although the specific mechanism of action is not known, macrolides are thought to exert their beneficial effects through anti-inflammatory rather than anti-microbial activities.
Post-transplantation Lymphoproliferative Disorder (PTLD)
PTLD occurs in approximately 1% of HSCT patients, usually within the first 4–12 months after transplantation [1, 64]. The clinical constellation may include fever, lymphadenopathy, pharyngitis, hepatosplenomegaly, and neurologic symptoms. There appears to be a greater incidence of fulminant, disseminated PTLD in HSCT recipients as compared to solid organ transplant recipients, possibly accounting for the increased mortality associated with PTLD in this population . The lung is involved only 20% of the time, usually as a component of disseminated disease, most commonly with ill-defined nodular infiltrates. It can also present as well-defined nodules, surrounded by a rim of ground-glass density (halo sign), mimicking the features of invasive aspergillosis. Hilar and mediastinal adenopathy and pleural effusions may also be seen. The pathogenesis of PTLD and its treatment are addressed below under Solid Organ Transplantation.
Solid Organ Transplantation
A number of non-infectious pulmonary complications affecting solid organ transplant recipients present with clinical and radiographic features that may mimic infection. These are described in the following sections; complications limited to specific organ recipient populations are noted in the subheadings.
Primary Graft Dysfunction (Lung Transplantation)
Primary graft dysfunction (PGD) represents a form of acute lung injury associated with the development of non-cardiogenic pulmonary edema within the first 72 h following lung transplantation . It is presumed to result from ischemia-reperfusion injury to the allograft(s), but inflammatory events triggered by brain death in the donor prior to implantation, as well as surgical trauma and lymphatic disruption, may be contributing factors. In most cases, the process is mild and transient, with fleeting pulmonary infiltrates on chest x-ray. In approximately 10% of cases, however, the presentation and course are similar to the acute respiratory distress syndrome (ARDS) with severe hypoxemia, widespread airspace opacities on chest x-ray, and the need for mechanical ventilator support. In common with ARDS, lung biopsies performed in patients with PGD demonstrate a prevailing pattern of diffuse alveolar damage.
A multitude of factors have been identified as associated with an increased risk of developing PGD, though the causal nature and mechanisms underlying these associations have not been established. Donor-related risk factors include female gender, African-American race, older age, and low donor PaO2/FiO2 ratio [67, 68, 69]. An elevated level of interleukin-8 in bronchoalveolar lavage fluid recovered from the donor has been associated with the development of severe PGD, supporting the notion that inflammatory events preceding organ harvest may play a role. Recipient risk factors include an underlying diagnosis of idiopathic pulmonary arterial hypertension as well as the presence of elevated pulmonary artery pressures independent of diagnosis. An association between graft ischemic time and PGD has not been consistently demonstrated. A possible explanation for the conflicting data is that ischemic time may become a factor only when it exceeds a certain threshold, suggested by one study as occurring beyond 6 h .
PGD should be considered when pulmonary infiltrates appear in the allograft(s) (sparing the native lung in cases of single lung transplantation) within the initial 3 days after lung transplantation. The diagnosis is one of exclusion. Other entities to be considered include volume overload, hyperacute rejection, aspiration pneumonitis, and pulmonary venous outflow obstruction. Pneumonia, transmitted from the donor via the allograft or acquired de novo post-transplantation, is an additional consideration. Evaluation should include assessment of hemodynamics (especially pulmonary capillary wedge pressure if a right heart catheter is in place), bronchoscopy to assess for purulent secretions and to obtain cultures, transesophageal echocardiography to visualize the pulmonary veins, and immunological testing for the presence of donor-specific anti-HLA antibodies.
As with ARDS due to other causes, treatment of severe PGD is supportive. Mechanical ventilation, employing a “low stretch” protocol, is the mainstay of care. Adjunct measures considered when oxygenation is tenuous include use of inhaled nitric oxide or prostacyclin and extracorporeal life support. Results of emergent retransplantation in this setting have been poor . Severe PGD is associated with a mortality rate in the range of 30–40% and represents a leading cause of perioperative death among lung transplant recipients. Recovery among survivors is often protracted, but achievement of normal graft function is possible. Survivors do appear to be at increased risk of developing bronchiolitis obliterans syndrome .
Allograft Rejection (Lung Transplantation)
Hyperacute rejection is a rare cause of widespread pulmonary infiltrates in the immediate postoperative period following lung transplantation . This form of rejection is mediated by preformed donor-specific anti-HLA antibodies present in the recipient at the time of transplantation. These antibodies target the pulmonary microvasculature, leading to complement- and neutrophil-mediated damage and widespread deposition of platelet/fibrin thrombi. Hyperacute rejection becomes clinically manifest within minutes to hours of establishing perfusion to the freshly implanted allograft. Intraoperatively, the allograft often appears dusky, mottled, and grossly edematous. Profound hypoxemia, hemodynamic instability, and dense opacification of the allograft(s) on chest x-ray are accompanying features. Four of five patients with this complication reported in the literature died; the one survivor was treated with a combination of plasmapheresis, anti-thymocyte globulin, and cyclophosphamide [74, 75]. Routine screening of all lung transplant candidates for preformed anti-HLA antibodies and either avoidance of donors with the targeted antigens or prospective cross-matching prior to transplantation have proven to be highly effective in minimizing the risk of hyperacute rejection.
Antibody-mediated rejection is a more recently recognized but still ill-defined form of acute rejection in lung transplant recipients . In contrast to hyperacute rejection, in which donor-specific anti-HLA alloantibodies are present in the recipient at the time of transplantation, this process is mediated by antibodies that develop de novo after transplantation, and it is therefore delayed in onset. The clinical presentation can be indistinguishable from acute cellular rejection or infection, with dyspnea, hypoxemia, and diffuse radiographic opacities. Hemoptysis, reflecting the presence of capillaritis, is an important clue but occurs in only 25% of cases . Proposed diagnostic criteria for acute antibody-mediated rejection are (1) presence of circulating donor-specific anti-HLA antibodies, (2) histopathological evidence of capillaritis, and (3) detection of endothelial cell C4d deposition. Treatment with high-dose corticosteroids is effective in less than half of patients; the addition of plasmapheresis is beneficial in the majority of steroid-refractory cases . Intravenous immunoglobulin and anti-CD20 monoclonal antibodies have also been used as adjunctive therapy.
Post-transplantation Lymphoproliferative Disorder (PTLD)
PTLD encompasses a spectrum of abnormal proliferative responses involving B cells in the majority of cases and ranging from benign hyperplasia to frank lymphomas. Epstein-Barr virus is responsible for driving B-cell proliferation in approximately 90% of cases. Proliferation occurs in an unregulated fashion due to absence of the normal cytotoxic T-cell response in the immunosuppressed patient. The proliferating B cells are of recipient origin in most cases. In contrast to B-cell-derived PTLD, the less commonly encountered T-cell neoplasms are predominantly EBV-negative.
The prevalence of PTLD varies considerably among the different solid organ transplant populations. The prevalence is lowest in kidney recipients (1%); intermediate in liver (2–5%), heart (2–5%), and lung (2–8%) transplant recipients; and highest in bowel transplant recipients (up to 30%) [80, 81]. Across all organ types, EBV-naïve recipients who receive organs from EBV-positive donors are at greatest risk for developing PTLD . The net state of immunosuppression and, in particular, the use of anti-lymphocyte antibodies, has also been implicated as a risk factor. A recent study of lung transplant recipients documented a decline in the incidence of PTLD at one large center in recent years; the authors speculate that this may relate to the shift from anti-lymphocyte antibodies to the less immunosuppressive interleukin-2 antagonists for induction .
Definitive diagnosis of PTLD requires tissue biopsy; fine needle aspiration rarely yields sufficient material to establish a diagnosis with confidence. Pathological analysis should include flow cytometry to determine clonality, in situ hybridization or immunohistochemical staining to assess for the presence of EBV, and determination of CD-20 expression to assist in planning treatment. Determination of EBV viral load by quantitative polymerase chain reaction assays has been touted as an ancillary diagnostic tool. However, this technique is limited by a lack of consensus on the appropriate specimen source (serum, whole blood, or peripheral mononuclear cells) and by varying threshold value definitions of a positive result. As a consequence of this, performance characteristics of EBV viral load testing in the diagnosis of PTLD vary considerably in the published literature [85, 86].
The initial treatment of PTLD involves reduction in the magnitude of immunosuppression to allow partial reconstitution of host T cellular immunity against EBV. Regression of tumor ensues in up to three-quarters of patients, typically within 2–4 weeks . While often successful, reduction in immunosuppression carries the attendant risk of precipitating acute or chronic allograft rejection, documented in 39% of patients in one series . Factors predictive of failure to respond to reduced immunosuppression include elevated serum lactate dehydrogenase level, severe organ dysfunction (need for hemodialysis, mechanical ventilation, vasopressors (bilirubin >4 mg/ dL), and multiple visceral sites of involvement .
For patients with CD-20-positive PTLD who fail to respond to reduced immunosuppression alone or have more aggressive tumors, administration of anti-CD20 monoclonal antibodies (rituximab) has emerged as the treatment of choice. This agent is generally well-tolerated and has been associated with remission rates of up to 60% and improved survival [88, 89]. Standard chemotherapy is reserved for patients with CD-20-negative PTLD, for rituximab failures, and for aggressive, life-threatening disease. While effective, chemotherapy is often poorly tolerated and associated with a significant risk of lethal infectious complications . Antiviral therapy is not effective in the treatment of established PTLD, but prophylactic use of antiviral agents for other purposes has been associated with a reduced risk of subsequent development of PTLD.
Lung Cancer (Heart and Lung Transplantation)
Sirolimus (mTOR Inhibitor) Pneumonitis
Multiple common disorders in transplantation patients are associated with radiographic lung infiltrates that can be confused with infectious pneumonia. While pneumonia is a serious complication in transplantation patients, leading to an appropriately high index of suspicion, accurate diagnosis of both infectious and non-infectious etiologies of lung infiltrates is essential to optimal treatment. The identification of non-infectious etiologies of lung infiltrates can usually be made on the basis of clinical findings and imaging studies, but invasive studies are sometimes necessary.
- 4.Hunt SA, Abraham WT, Chin MH, Feldman AM, Francis GS, Ganiats TG, et al. 2009 focused update incorporated into the ACC/AHA 2005 guidelines for the diagnosis and management of heart failure in adults: a report of the American College of Cardiology Foundation/American Heart Association Task Force on Practice Guidelines: developed in collaboration with the International Society for Heart and Lung Transplantation. Circulation. 2009;119(14):e391–479.PubMedPubMedCentralGoogle Scholar
- 5.Wilson JF. In the clinic. Heart failure. Ann Intern Med. 2007;147(11):ITC12-1-ITC-6.Google Scholar
- 19.Yanik GA, Ho VT, Levine JE, White ES, Braun T, Antin JH, et al. The impact of soluble tumor necrosis factor receptor etanercept on the treatment of idiopathic pneumonia syndrome after allogeneic hematopoietic stem cell transplantation. Blood. 2008;112(8):3073–81.PubMedPubMedCentralCrossRefGoogle Scholar
- 37.Moskowitz CH, Bertino JR, Glassman JR, Hedrick EE, Hunte S, Coady-Lyons N, et al. Ifosfamide, carboplatin, and etoposide: a highly effective cytoreduction and peripheral-blood progenitor-cell mobilization regimen for transplant-eligible patients with non-Hodgkin’s lymphoma. J Clin Oncol. 1999;17(12):3776–85.PubMedCrossRefPubMedCentralGoogle Scholar
- 38.Di Ianni M, Ballanti S, Iodice G, Reale A, Falzetti F, Minelli O, et al. High-dose thiotepa, etoposide and carboplatin as conditioning regimen for autologous stem cell transplantation in patients with high-risk Hodgkin’s lymphoma. Hematology. 2012;17(1):23–7.PubMedCrossRefPubMedCentralGoogle Scholar
- 42.Sirvent-von Bueltzingsloewen A, Gratecos N, Legros L, Fuzibet JG, Cassuto JP. Antithymocyte-associated reticulonodular pneumonitis during a conditioning regimen of reduced intensity for genoidentical bone marrow transplantation. Bone Marrow Transplant. 2001;27(8):891–2.PubMedCrossRefPubMedCentralGoogle Scholar
- 87.Tsai DE, Hardy CL, Tomaszewski JE, Kotloff RM, Oltoff KM, Somer BG, et al. Reduction in immunosuppression as initial therapy for posttransplant lymphoproliferative disorder: analysis of prognostic variables and long-term follow-up of 42 adult patients. Transplantation. 2001;71(8):1076–88.PubMedCrossRefPubMedCentralGoogle Scholar
- 88.Evens AM, David KA, Helenowski I, Nelson B, Kaufman D, Kircher SM, et al. Multicenter analysis of 80 solid organ transplantation recipients with post-transplantation lymphoproliferative disease: outcomes and prognostic factors in the modern era. J Clin Oncol. 2010;28(6):1038–46.PubMedPubMedCentralCrossRefGoogle Scholar