RESEARCH CORRESPONDENCE Diagnosis of hemolysis and device thrombosis with lactate dehydrogenase during left ventricular assist device support Palak Shah, MD, MS, a Vivek M. Mehta, BS, b Jennifer A. Cowger, MD, a Keith D. Aaronson, MD, MS, a and Francis D. Pagani, MD, PhD c From the a Division of Cardiovascular Medicine; b College of Literature, Science and Arts; and the c Department of Cardiac Surgery, University of Michigan, Ann Arbor, Michigan Mechanical circulatory support (MCS) with a continuous- flow (CF) rotary pump is associated with low levels of hemolysis during normal pump operation. Clinically significant hemolysis can be an early sequelae of pump thrombus. Left ventricular assist device (LVAD) thrombosis may result in stroke, peripheral embolism, heart failure, device exchange, and death. 1,2 The timely detection of pump thrombus remains challenging. Serum free hemoglobin (sfHg) is the most widely accepted marker identifying LVAD hemolysis. A sfHg 4 40 mg/dl with accompanying clinical signs and symptoms has been used to define significant hemolysis in the Interagency Registry of Mechanically Assisted Circulatory Support (INTERMACS). 3 The use of lactate dehydrogenase (LDH) has also been advocated for detection of significant hemolysis. 4 It is unknown what threshold of LDH value should be set to differentiate low levels of hemolysis reflecting normal pump operation from abnormal levels of hemolysis reflecting device thrombosis. The objective of our study was to define values of LDH and sfHg consistent with normal pump operation and to determine thresholds signaling concern for device thrombosis. Our report consists of all patients who had primary implantation of a CF LVAD at the University of Michigan Health System from 2003 to 2012. Patients were supported on CF axial design (CF-A) and CF centrifugal design (CF-C) devices. Detailed methods are available on the jhltonline.org Web site. Criteria for clinical suspicion of device thrombosis included: (1) hemoglobinuria or hyperbilirubinemia; (2) heart failure not explained by right ventricular failure or valvular heart disease (eg, aortic insufficiency); (3) hospi- tal readmission for suspected hemolysis or device throm- bosis; and (4) sustained power elevations or abnormal device function. Confirmation of device thrombosis oc- curred by manufacturer analysis of the explanted device or by direct visualization at the time of explant. Patients who did not fulfill any of the listed criteria for suspected or confirmed device thrombosis were identified as a control population. No device analyses were available for patients who underwent CF-C device exchange. Hemolysis markers were only evaluated in CF-C devices from the control population that had no clinical suspicion of device thrombosis. During the study period, 254 patients underwent primary implantation of a CF LVAD, with 241 included in the final analysis. Baseline demographics of the cohort are reported in Supplemental Table 1 (available on the jhltonline.org Web site). There were 196 CF-A (81%) and 45 CF-C LVADs (19%). The CF-A group comprised 144 controls (73%), 21 patients (11%) with confirmed device thrombosis, and 31 patients (16%) with clinically suspected but unconfirmed device thrombosis. The CF-C group comprised 40 controls (89%) and 5 patients (11%) with suspected but unconfirmed device thrombosis. LDH values within the control patients were lower for those supported by CF-C LVADs (254  46 IU/L) vs CF-A LVADs (361  76 IU/L, p o 0.001; Supplemental Table 2, available on the jhltonline.org Web site). There was no difference in sfHg between the CF-A (10.6  5.3 mg/dl) and CF-C groups (10.1  4.4 mg/dl, p ¼ 0.542; Supplemental Table 2, available on the jhltonline.org Web site). To validate LDH values in the control population, 10 patients with a CF-A device who underwent device exchange for infection, with absence of device thrombus on manufacturer analysis, were identified. These patients had a mean LDH of 359  89 IU/L (p ¼ 0.958), which was similar to the remainder of the CF-A LVAD control population. Hemolysis markers in the setting of device thrombosis were assessed in 21 CF-A patients. Mean values for LDH (923  560 IU/L) and sfHg (26.8  18.8 mg/dl) were significantly higher for patients with confirmed device thrombosis than for CF-A controls (p o 0.001 for both comparisons; Supplemental Table 2, available on the jhltonline.org Web site). The area under the curve for LDH as a marker of device thrombosis was 0.94  0.01, which was significantly higher than the 0.79  0.01 calculated for sfHg (p o 0.001, Figure 1). 5,6 A sfHg threshold of 40 mg/dl had a sensitivity of 25% and specificity of 98% for detection of device thrombosis (Supplemental Table 3, available on the jhlton- line.org Web site). A threshold of 600 IU/L was selected for LDH, representing the mean plus 3 standard deviations of the values in the control population. Assuming a normal distribution, this threshold would exclude more than 99% 1053-2498/$ - see front matter r 2014 International Society for Heart and Lung Transplantation. All rights reserved. http://dx.doi.org/10.1016/j.healun.2013.07.013