Keywords

Thrombocytopenia

Sepsis accounts for 10–20% of ICU admissions [1,2,3,4], and the incidence of thrombocytopenia in sepsis ranges between 43.2% and 70.6% [5,6,7]. Though thrombocytopenia is a common laboratory finding for critically ill patients and indicates poor prognosis [8,9,10,11,12,13,14,15,16], there are still many unresolved controversies regarding the management of thrombocytopenia in critically ill patients.

The Incidence and Outcome of Thrombocytopenia in ICU

At the time of ICU admission, approximately 50% of patients, ranging between 8.3% and 67.6%, have thrombocytopenia, while 14–44% of patients develop thrombocytopenia during their ICU stay [16, 17]. The cutoff values chosen for thrombocytopenia, differences in patient populations, and ICU characteristics are considered to account for the large variation between series.

From the clinical perspective, in most cases the development of thrombocytopenia in the ICU indicates serious organ system derangement and physiologic decompensation rather than a primary hematologic disorder. Multivariate analysis identified risk factors associated with ICU-acquired thrombocytopenia as severity of illness, organ dysfunction, sepsis, vasopressor use, and renal failure [16,17,18]. The presence of thrombocytopenia and its severity, especially severe thrombocytopenia with platelet counts below 50 × 109/L, are associated with bleeding, increased risk of transfusion (red cells, platelets, plasma, or cryoprecipitate) [17, 19], and adverse clinical outcomes including death, though few deaths are directly attributable to bleeding [17,18,19,20,21]. Multivariate analysis of six observational studies found thrombocytopenia to be an independent predictor of mortality [17]. In the PROTECT trial, both moderate and severe thrombocytopenias were associated with increased ICU and hospital length of stay as well as the need for organ support [19].

The recovery rate of thrombocytopenia further predicts outcome; a blunted platelet count recovery is associated with increased mortality risk [22], while rapid recovery suggests survival [13]. A large prospective study has shown that the platelet counts in sepsis drop rapidly after ICU admission and usually reach their nadir on day 4 [21]. By day 7 the platelet count recovers to the admission value and by day 9 exceeds the admission value [21]. Non-survivors were much more likely to not experience platelet recovery [8, 21]. These observations suggest that if the platelet count decreases after partial recovery in a patient admitted for sepsis, the worsening thrombocytopenia should not be blamed on the initial infection. Instead, alternative causes such as HIT, drug-induced thrombocytopenia, dilution, or a secondary infection should be sought [6, 20, 21, 23, 24].

Thrombopoiesis and Mechanisms of Thrombocytopenia in Sepsis

Peripheral blood platelet counts are tightly controlled by interactions regulating platelet production from the bone marrow, pooling in the liver and spleen, and elimination in the reticuloendothelial system [25, 26]. Normal platelet count in humans ranges from 150 to 400 × 109/L, and platelets have a circulating life span of approximately 10 days. About one-third of platelets are sequestered in the spleen at any time. To maintain a normal platelet count, approximately 100 × 109 platelets are released from mature megakaryocytes into circulation each day [25]. Several mechanisms that contribute to thrombocytopenia including hemodilution, platelet consumption, decreased platelet production, increased sequestration of platelets, immune-mediated destruction, and pseudo-thrombocytopenia [16, 20, 23] are summarized in Table 7.1. In the ICU setting, often more than one mechanism is responsible for low platelet counts.

Table 7.1 Mechanisms of thrombocytopenia in the ICU
Full size table

Causes of thrombocytopenia in the septic patient are often multifactorial including consumption, immune-mediated thrombocytopenia, and suppression of platelet production. Additionally, critical care patients frequently have many comorbid illnesses that decrease platelet counts through platelet sequestration or drug-induced thrombocytopenia including the HIT syndrome. Of these mechanisms, consumptive coagulopathy is probably the most significant cause of thrombocytopenia in the septic patient.

Suppression of platelet production is frequently cited as a cause of thrombocytopenia in septic patients; however inflammatory cytokines are known to blunt erythropoiesis but also usually stimulate thrombopoiesis. In a small series, bone marrow aspirates of septic patients revealed normal numbers of megakaryocytes [27]. Circulating levels of thrombopoietin are often elevated in sepsis [28], and numbers of reticulated platelets have been shown to be increased in septic neonates with thrombocytopenia. Therefore, the data so far is not convincing for direct megakaryopoiesis inhibition. Relative insufficiency due to increased consumption is more plausible [7, 29].

Increasingly platelets are recognized to have a role as immune effectors in addition to their well-known role in hemostasis [30], and thrombocytopenia has been associated with a dysregulated host immune response in sepsis patients [31].

In the normal physiologic state, coagulation is finely regulated between procoagulant and anticoagulant pathways. Sepsis results in dysregulation of these two pathways, overwhelming activation of the procoagulant pathway, suppression of the anticoagulant pathway, and widespread fibrin deposition throughout the microvasculature resulting in further organ dysfunction [32,33,34]. Paradoxically, this procoagulant state can also increase bleeding risk as procoagulant factors, fibrinogen, and platelets are consumed [34]. Disseminated intravascular coagulation (DIC) is a final common syndrome, which, in sepsis, is caused by a heterogeneous group of infectious insults that result in microvascular thrombin deposition in numerous organs and multiple organ dysfunction. Even in the absence of overt DIC, coagulation activation markers, i.e., thrombin-antithrombin complexes and prothrombin fragment 1.2, are increased in sepsis. Though there has been enough evidence that sepsis-induced endothelial damage activates procoagulant pathways, clinical trials that evaluated treatment with anticoagulant factors, i.e., tissue factor pathway inhibitor, antithrombin, and activated protein C, had negative or mixed results [34,35,36,37].

Hemophagocytic lymphohistiocytosis (HLH) or macrophage activation syndrome has been increasingly reported in patients with thrombocytopenia and sepsis [38]. Treatment options for this diagnosis include dexamethasone, cyclosporine, and etoposide.

In animal studies, during cellular injury in sepsis, extracellular histones are released, and histones induce platelet aggregation causing a drop in platelet counts [39]. Clinical studies have shown correlation between histone plasma levels and thrombocytopenia [40], but the validity of this correlation in the clinical setting requires further evaluation.

Although not as severe as TTP [41], ADAMTS 13 levels are shown to be reduced in sepsis patients and may contribute to ongoing microcirculation insufficiency, thrombosis, and organ failure.

Immune complex and bacterial wall exposure activate complement pathways during sepsis. Complements bind and activate platelets therefore contributing to platelet consumption [20].

Evaluation of Thrombocytopenia in Shock and Sepsis

Evaluation of thrombocytopenia in critically ill septic patient requires a stepwise approach. Prior medical history, comorbidities, detailed drug or toxin exposure history, and prior blood count results provide invaluable information. Prior blood count results can reveal preceding bone marrow disorders such as myelodysplasia, chronic ITP, and lymphoproliferative disorders. The course of disease presentation, i.e., acute onset, gradual decline, and acute decompensation, over known chronic medical problems should be taken into account for management planning.

The course of thrombocytopenia during the hospital stay also defines treatment and management.

Patients that experience rapid platelet count decreases in a few days and then rapidly recover after improvement of critical illness usually have acute multisystem inflammatory process, i.e., burns, pancreatitis, or sepsis. In sepsis, platelet count nadir happens within 2 days of presentation, and median recovery takes about 6 days. Median time from vasopressor discontinuation to platelet count recovery above 100 × 109/L is approximately 2 days.

On the other hand, major trauma, surgery, massive transfusion, or cardiopulmonary bypass usually induces rapid, immediate platelet count decrease, and if the process is not complicated, patients experience rapid recovery of their platelet counts. Platelet counts start to increase 3–4 days after surgery and should continue to improve over the next 1–2 weeks. As thrombopoietic factors, i.e., thrombopoietin, IL-6, and acute-phase cytokines, are released, patients can have platelet counts above their baseline through recovery [8, 14, 21, 23, 24, 42].

Patients who experience low platelet counts at presentation and throughout the duration of the ICU stay should be considered to have an underlying bone marrow process or hypersplenism, and as such, thrombocytopenia evaluation should include imaging studies as well as bone marrow aspiration and biopsy.

The incidence of chemotherapy-induced thrombocytopenia varies greatly depending on the treatment used; the highest rates of this condition are associated with gemcitabine- and platinum-based regimens [43]. Thrombocytopenia due to cytotoxic chemotherapy usually occurs between 1 and 2 weeks after treatment and then recovers rapidly [44]. Chemotherapeutic agents such as gemcitabine can induce thrombotic thrombocytopenic purpura or hemolytic uremic syndrome [45, 46].

Patients, who develop gradually worsening thrombocytopenia throughout their ICU stay, occasionally as their clinical course is improving, should be evaluated for iatrogenic thrombocytopenia, i.e., drug-related immune thrombocytopenia, heparin-induced thrombocytopenia, or posttransfusion purpura. In these cases, early recognition and testing for etiology provide definitive treatment options, i.e., removal of the offending drug, intravenous immunoglobulin, and change in anticoagulation.

Patients who develop thrombocytopenia rapidly within a few days of ICU admission and continue to have moderate-to-severe thrombocytopenia with persistent multiorgan failure have the worst prognosis. These patients traditionally have persistent thrombocytopenia as an indicator of the severity of the underlying disease such as sepsis, DIC, or shock.

Severe thrombocytopenia, defined by a platelet count nadir below 20 × 109/L, has a limited differential diagnosis and most commonly suggests consumptive thrombocytopenia due to DIC, bone marrow failure or immune-mediated mechanisms.

Evaluation of the coagulation cascade aids in the diagnosis of consumptive thrombocytopenia due to DIC. High-fibrin degradation products and D-dimer, low fibrinogen, prolonged prothrombin time, and partial thromboplastin time would be noted depending on the severity of DIC.

Bone marrow failure syndromes either due to infiltration, myelosuppressive treatment, or hemophagocytosis also present with severe thrombocytopenia, and usually other hematopoietic elements are affected, i.e., neutropenia, anemia, and reticulocytopenia.

Shortened platelet survival due to immune-mediated mechanisms, such as autoimmune ITP or alloimmune disorders due to HLA antibodies, posttransfusion purpura, or drug related, usually cause prominent severe thrombocytopenia. HIT is the exception that usually does not cause severe thrombocytopenia.

Thrombosis in the setting of thrombocytopenia usually occurs due to DIC, HIT, or anti-phospholipid syndrome. In the setting of coagulation cascade activation, natural anticoagulants may deplete faster when patients have underlying liver disease, or acute ischemic liver injury, which results in peripheral small vessel thrombosis and gangrene despite adequate arterial flow [47].

Management of Thrombocytopenia in Shock and Sepsis

Most patients who are critically ill present with thrombocytopenia as a part of their underlying disease. Understanding the physiological changes associated with disease can help to define the projected course of platelet recovery and timely investigation and treatment when platelet recovery does not follow the anticipated course.

The principal treatment of ICU-associated thrombocytopenia is to treat the underlying cause. Appropriate antimicrobial therapy, source control, and organ support in patients with sepsis are the essential steps of treating thrombocytopenia [20, 21, 24, 48]. Hemostatic treatment by transfusion and coagulation support, i.e., cryoprecipitate, fresh frozen plasma, and platelets, should be considered for patients with bleeding symptoms. Available data is not clear for prophylactic transfusion in critically ill patients, unless the underlying mechanism for thrombocytopenia is due to major surgery or trauma. The American Association of Blood Banks and the Society of Critical Care 2016 Surviving Sepsis Guidelines recommend prophylactic platelet transfusion below a threshold of 10 × 109/L [48,49,50]. Unfortunately, these recommendations are extrapolated from randomized trials evaluating prophylactic platelet transfusion in adult patients who developed thrombocytopenia due to chemotherapy exposure. In the setting of critically ill patients, platelet consumption and platelet dysfunction also contribute to bleeding risk [6, 7, 34]. Platelet dysfunction is usually aggravated by factors including uremia, liver failure, antiplatelet agents, and extracorporeal circuits. Additionally, response to platelet transfusion in terms of the posttransfusion platelet increment and the durability of platelet increase are known to be altered in critically ill patients [51]. Although the results vary considerably across patients, one-unit platelet transfusion results in a median increase in the platelet count of 15 × 109/L [51]. Correction of thrombocytopenia should always be pursued when there is a significant risk of bleeding; however, major hemorrhage can occur at any platelet count as a consequence of multiple organ failure, and transfusion associated risks should always be considered. Platelet transfusions are associated with increased risk of nosocomial infection and transfusion associated lung injury [49, 52, 53].

Upregulation of tissue factor expression on monocytes, and other cells in response to inflammatory mediators, i.e., endotoxin, triggers coagulation cascade activation in sepsis and DIC. Inhibition of coagulation cascade with heparin might have a role in breaking the consumption circuit in DIC and might avoid further thrombocyte consumption. Randomized trials investigating the role of heparin in sepsis indicate some clinical outcome improvement [54]. Though this meta-analysis suggests decreased mortality in patients with sepsis, septic shock, and disseminated intravascular coagulation, the overall impact remains uncertain [54]. Further multicenter clinical trials are investigating the safety outcomes of heparin use in sepsis.

Transfusion support in critically ill patients relies on physician judgement rather than data, and the effectiveness of platelet transfusion to reduce transfusion of other blood components, or improve overall outcome is obscure [17, 18, 51]. We are still in need of large randomized trials evaluating transfusion strategies to treat or prevent bleeding in critically ill thrombocytopenic patients. The guidelines for platelet transfusion summarized in Table 7.2 [6, 26, 49, 51, 55] depend on either low or very low quality of evidence.

Table 7.2 Platelet transfusion recommendations in critically ill patient [21, 49,50,51]
Full size table

Therapeutic platelet transfusions are indicated in patients with bleeding who have more than mild blood loss like epistaxis, hematuria, and hematemesis, at or above bleeding World Health Organization Grade 2. In ICU patients thrombocytopenia is also associated with acquired mild-to-moderate platelet dysfunction [56,57,58] caused or aggravated by medications, i.e., antibiotics [59, 60] and analgesics, activation of platelets on extracorporeal circuits [61], or cleavage of platelet receptors by enzymes released in sepsis [7]. Therefore bleeding symptoms are much more relevant for transfusion decision-making than the actual platelet count. In patients who may have immune thrombocytopenia, drug-induced immune thrombocytopenia, or posttransfusion purpura, platelet transfusion should only be considered for serious or life-threatening bleeding, as the survival of transfused platelets is remarkably low in patients who have ongoing immune-mediated platelet destruction. Platelet transfusions should be avoided in patients who have low platelet counts with thrombosis, i.e., HIT [62] and TTP [63], as that might push the balance toward thrombosis. Platelet transfusions are considered to be useful in patients with platelet loss or consumption but might cause serious adverse events in patients with increased intravascular platelet activation, i.e., HIT or TTP. Therefore, successful treatment requires identifying the underlying cause of thrombocytopenia.

In patients with hypoproliferative thrombocytopenia due to underlying bone marrow failure or chemotherapy exposure, but without major bleeding symptoms, prophylactic platelet transfusion is based on clinical judgment. Critically ill thrombocytopenic patients who have nonimmune thrombocytopenia and spontaneous oropharyngeal mucous membrane bleeding have increased risk of bleeding into the central nervous system or retina and should receive platelet transfusion. For patients who have platelet counts above 10 × 109/L, prophylactic platelet transfusion should be considered for patients with active bleeding risks or before invasive procedures (Table 7.2).

Platelet count increment needs to be calculated after platelet transfusions to distinguish platelet consumption versus refractoriness, as multiparous women and patients with a history of multiple transfusions could have anti-HLA antibodies, necessitating HLA-matched platelet transfusion. A unit apheresis platelet (or a pool of six platelet concentrates) should achieve an increment of 30,000–50,000/μL in an average adult; however, for critically ill patients, expected mean platelet count increase is approximately 15 × 109/L. Checking a platelet count 15–30 min after platelet transfusion, and calculating corrected count increment, provides valuable information if the patient should be further evaluated for platelet-reactive antibodies, i.e., anti-HLA class I antibodies or rarely antihuman platelet antigen antibodies. Corrected count increment or CCI is usually calculated with the following formula: CCI = (platelet count posttransfusion − platelet count pretransfusion) × BSA (m2)/number of platelets transfused × (1011). CCI below 5500–7000 is considered an inadequate response. Patients with platelet transfusion refractoriness require a multidisciplinary approach, i.e., hematology and blood bank, to manage their need for HLA-matched platelets.

Management of thrombocytopenia in the critically ill is still often determined by expert opinion, and the guidelines summarized in Table 7.2 serve as suggestions until we have definitive data.

White Blood Cell Alterations in Sepsis

Increased production and activation of neutrophils and monocytes are important in the host’s response to infection [64]. In animals, intravenous injection of endotoxin produces a marked increase in number of circulating leukocytes [65]. In patients with sepsis, the white blood cell count rises, i.e., leukocytosis. Neutrophils play important roles in fighting infection through phagocytosis of infecting microorganisms as well as immune complexes [66]. Neutrophilia is produced by a combination of de-margination, increased bone marrow release, and increased production of neutrophils. When bone marrow reserves are mobilized, there is an increased number of premature band forms released into the circulation producing a “left shift” in the peripheral blood white cell count. Although a neutrophilic leukocytosis is the appropriate response to many types of infections, neutropenia may develop in some patients and is associated with a poor prognosis [67,68,69]. This may represent an overwhelming immune suppression of marrow. Sepsis is also associated with activation of monocytes. The monocytes/macrophage is the human equivalent of the invertebrate amoebocyte, the innate immune cell responsible for reaction to pathogens as well as wound repair. The ability of the monocyte to upregulate synthesis and release of cytokines as well as expression of tissue factor provides critical cross-talk between inflammation and coagulation. Monocytes possess toll-like receptors that are a key component of the innate immune system. These pathogen-associated molecular pattern signaling receptors are a key to appropriate immune response [70].

Leukemoid reaction, defined as leukocyte counts in excess of 50,000, is an extreme response to infection.

Since the white blood cell is a key effector response to counter bacterial challenge, it is appealing to postulate that granulocyte colony-stimulating factor (G-CSF) would improve outcome in severe infection. Although theoretically attractive, especially in patients with sepsis and neutropenia, there is no evidence that this approach improves outcomes.

Immature granulocytes would be expected to differentiate a systemic inflammatory response related to infection versus one that is not. It has been shown that the immature granulocyte count does discriminate between infected and non-infected patients [71]. Although a differentiator of the presence or absence of an infection, the immature granulocyte count does not serve as a prognostic marker for mortality. It may, however, be useful as a rapid screening tool in patients with potential infection. The “delta neutrophil index” measures the immature granulocyte fraction with an automated blood cell analyzer to create a delta that represents the difference between the mature PMN leucocytes and the total leucocyte count [72]. The delta neutrophil index expressed as a percentage has been shown to differentiate between controls, a systemic inflammatory response (SIRS) without infection, SIRS with infection, and severe sepsis (infection-induced organ dysfunction) better than C-reactive protein (CRP).

CD64 is an antigen expressed on neutrophils during an infectious or inflammatory state. In a bacterial infection, there is an increase in the density of CD64 antigen on neutrophils and an increase in the number of neutrophils expressing CD64. This is not typical for viral infections. CD64 expression may, however, also be seen in noninfectious inflammatory conditions such as sickle cell crisis or acute bouts of rheumatoid arthritis. Similar to the delta neutrophil index discussed above, the percentage of CD64-positive neutrophils is highest in patients with sepsis, followed by SIRS in the absence of infection and lowest in normal controls [73].

It is now recognized that in some patients the pro-inflammatory response to severe sepsis is replaced over time with an anti-inflammatory response that can be identified by the presence of persistent lymphopenia [74]. This immunosuppressive phase is characterized by apoptosis-related loss of immune cells including CD4, CD8, and follicular dendritic cells [75, 76]. It is possible that pro-immune therapy would be anti-apoptotic, prevent lymphocyte death, and improve survival in severe sepsis. Animal data would support this possibility and indicate that immunomodulatory therapy offers promise in future clinical research [77, 78]. Absolute lymphocyte counts are easily measured during routine care.

Red Blood Cell and Blood Transfusion

Red Blood Cell Rheology in Sepsis

Alterations in RBC rheology may contribute to the microvascular injury and impaired oxygen supply seen in sepsis.

Major determinants of red blood cell (RBC) rheology are viscosity, aggregation, deformability, and membrane physiology (X1). Viscosity is the extent to which a fluid resists flow. In very technical terms, it is the tangential stress on the flow of a liquid divided by the velocity gradient. It relates to internal friction defined by force per unit area of flow resistance. The major determinants of whole-blood viscosity are plasma viscosity, hematocrit, and RBC deformability and aggregation.

Shear rate signifies the rate of change in velocity as one layer of fluid passes over an adjacent layer. Rouleaux are stacks of red blood cells (RBCs) which tend to form because of the unique discoid shape of the red blood cell. Formation of rouleaux increases blood viscosity. Low shear rates, high hematocrits, and low-flow states promote rouleaux formation. In the presence of inflammatory acute-phase proteins, rouleaux of cells drive aggregation of RBCs. Erythrocyte sedimentation rate estimates aggregation tendency in RBCs.

Deformability characterizes the RBC’s ability to undergo deformation during flow [79, 80]. RBC deformation when exposed to fluid forces is determined by membrane material properties cell geometry and cytoplasmic viscosity [81]. The RBC must deform to pass through the smaller capillaries and achieve oxygen transfer. Nitric oxide (NO) modulates RBC membrane properties and is released in abundance in severe sepsis and septic shock. NO, perhaps by an effect on the membrane calcium-ATPase channel, may decrease RBC deformability. Sepsis-induced production of reactive oxygen species has also been linked to decreased RBC deformability [82, 83].

Red Blood Cell Transfusion in Sepsis Hypoperfusion

In 1999 a large prospective randomized study of 838 stable intensive care unit patients assigned either to a hemoglobin target of 7–9 versus 10–12 g/dL was carried out [84]. Overall 30-day mortality was similar in the two groups with outcome better in the restrictive group and complications more frequent in the liberal group. Based on our personal observations at our own institution, there has been a steady increase in utilizing the 7 g/dL transfusion threshold in ICU patients in general.

A separate subgroup of ICU patients that deserves separate attention are those with septic shock or sepsis-induced hypoperfusion as manifested by lactate acidosis. Red blood cell transfusion has been shown to be a frequent intervention in patients with septic shock and anemia in the ICU [85,86,87,88]. Those who use a low threshold for transfusing patients with septic shock rationalize that blood transfusion would increase oxygen delivery (a given) and lead to increase oxygen consumption in those patients with an oxygen deficit due to tissue hypoperfusion. This is clearly controversial with some studies showing failure to increase oxygen consumption in septic patients with tissue hypoperfusion despite the increase in oxygen delivery, whereas others show increased oxygen consumption following blood transfusion, particularly in patients with elevated lactate or low superior vena cava O2 saturations [89,90,91,92]. Marik and Sibbald studied 23 critically ill patients with sepsis and failed to demonstrate a beneficial effect of red blood cell transfusion on measured systemic oxygen uptake [89]. Fernandes et al. studied 15 critically ill septic patients with Hgb levels <10 g/dL. Ten patients received one unit of blood, and five controls received albumin in the same volume. Increase in regional or global oxygen utilization could not be demonstrated [90]. Mazza and colleagues studied 46 patients with 74 transfusions and were able to show an association between blood transfusion and increase in ScvO2 and decrease in lactate [91]. In a surgical sepsis population, Steffles, Bender, and Levison were able to demonstrate augmentation of both DO2 and VO2 that could not be predicted by increased lactate levels. They postulated that patients with increased lactate levels may have peripheral oxygen utilization deficit that handicap improvement in VO2 with increase in DO2 [92]. One of the issues with transfusion of red blood cells relates to reduction in quality of the RBC during storage. During storage, changes in RBC morphology, changes in membrane surface characteristics, and decrease in RBC deformability are problematic [93].

Some have advocated transfusion of fresh blood as opposed to stored blood to avoid the issues elucidated above, but no data exists to support that advantage. One database study using propensity-matched analysis in patients with severe sepsis and septic shock did show an “association” between red blood cell transfusions and lower mortality [94]. Although the literature is conflicting as to the ability of increasing oxygen delivery to increase oxygen consumption, the predominance of high-quality literature and in particular the Holst study (see discussion to follow) showing no clinical outcome benefit of higher transfusion threshold in septic shock patients points to a preference for a more conservative transfusion in patients with severe sepsis and septic shock regardless of presence or absence of tissue hypoperfusion [95]. Although transfusion of RBCs would seem theoretically to be a potential solution for increasing tissue oxygen delivery, changes in septic patients’ RBCs induced by endogenous factors in the septic patients as well as effects of storage may ameliorate any potential positive effects in the microcirculation.

Blood Transfusion as Part of Early Goal-Directed Therapy for Septic Shock

In late 2001 Rivers et al. published a landmark paper, landmark in that it was a wake-up call as to our care of the septic shock patient. It was entitled “Early Goal-Directed Therapy in the Treatment of Severe Sepsis and Septic Shock” [96]. Over 260 patients in the emergency department (ED) with infection-induced hypotension (SBP < 90 mmHg after ≥20 cc/kg crystalloid IV bolus) and/or lactate ≥4 mmol/L were randomized to receive standard protocolized therapy or “early goal-directed therapy” (EGDT) as termed by the authors. Patients in both groups had central venous catheters placed and fluid resuscitation targeting a CVP of 8–12 mmHg, vasopressors titrated to a mean arterial pressure (MAP) of ≥65 mmHg, and a urine output of at least 0.5 cc/kg/h. The EGDT group had the same care as the control group plus therapy to raise the superior vena cava venous O2 saturation (ScvO2) to ≥70%, if necessary using dobutamine and/or blood. The in-hospital mortality was 30.5% in the EGDT group vs. 46.5% in the standard therapy group (p = 0.009). EGDT became part of many multifaceted sepsis performance improvement initiatives that demonstrated improved survival when compared to historical controls [97].

Surviving Sepsis Campaign Guidelines, Early Goal-Directed Therapy, ScvO2, and Blood Transfusion

The first surviving sepsis guidelines (SSC) were published in 2004 [98] and recommended that during the first 6 h of resuscitation in patients with sepsis-induced hypoperfusion (hypotension and/or lactic acidosis), resuscitation should target ScvO2 target of 70% using the EGDT methods of the Rivers trial. A mixed venous oxygen saturation (SvO2) of 65% or greater was offered as an alternative if a pulmonary artery catheter was present [99]. If fluid resuscitation, vasopressors as needed to achieve a MAP ≥ 65 mmHg, and normal oxygen saturation failed to raise the ScvO2 to 70%, then a transfusion of packed RBCs to raise the hematocrit to 30% and/or empiric dobutamine to raise the cardiac output was recommended in attempts to achieve that goal. The first revision of SSC guidelines occurred in 2008 [100] and again recommended the previous EGDT-delineated targets of CVP, MAP, urine output (UOP), and ScvO2. “Protocolized resuscitation” was used to describe the resuscitation recommendation, and the group to receive this resuscitation was specifically defined as persistent hypotension after fluid challenge and/or blood lactate concentration ≥4 mmol/L. This brought the SSC tissue hypoperfusion definitions in line with the original Rivers study. In the 2012 SSC guidelines [101], the term “protocolized quantitative resuscitation” was used to describe the recommended treatment of septic patients with tissue hypoperfusion, keeping the targets the same as in 2008. Lactate clearance was suggested as an additional target after studies by Jones et al. [102] showed that achieving 10% lactate clearance during resuscitation of sepsis-induced tissue hypoperfusion was equivalent to targeting an ScvO2 of >70%. The recommendations for transfusion of blood or dobutamine were removed from the guidelines with the physician charged to normalize ScvO2 by whatever method they chose. The limitations for the use of CVP as a marker of intravascular volume status and fluid responsiveness were recognized, and dynamic targets were advised as a consideration for determination of volume responsiveness.

Following the publication of the ARISE, PRoCESS, and PROMISE trials that showed measurement of CVP and ScvO2 was not necessary for the successful resuscitation of septic shock [103,104,105]. The Centers for Medicare and Medicaid Services (CMS) published the CMS SEP-1 measures which did not require targeting or even measurement CVP and ScvO2, but these variables were still one option for satisfying the reevaluation of intravascular volume and tissue perfusion following initial fluid resuscitation. If ScvO2 is chosen as a target for resuscitation of septic shock, what therapeutic interventions, if any, are needed beyond intravascular volume expansion and vasopressor agents? In septic shock patients, Jones et al., after aggressive administration of intravenous crystalloid to achieve a CVP of 8–12 mmHg and vasopressor agents to achieve an arterial pressure of 65 mmHg or greater in a stepwise resuscitation algorithm, found that very few patients were given packed red blood cell transfusion in an attempt to achieve ScvO2 goals (only 7% received packed red blood cell transfusion) [102]. In contrast, 64% of patients in the EGDT (ScvO2 targeted) arm in the Rivers study were treated with blood transfusions. The Rivers data suggested that packed red blood cell transfusions were part of the success of the EGDT arm, whereas the Jones trial (and other observational clinical studies) suggests that these additional therapies are rarely required to achieve ScvO2 goal, if utilized as a target.

The 2016 revision of the SSC guidelines has no recommendation for specific targets of resuscitation and emphasizes the importance of septic shock as a medical emergency with initial 30 mL/kg crystalloid to be given within 3 h of presentation with additional fluids guided by reassessment and the suggestion of dynamic (pulse pressure variation, stroke volume variation) over static (CVP measurement) to predict fluid responsiveness [106]. Although not part of a recommendation, the body of the document does point to the fact that there were equal outcomes in the ScvO2 targeted and usual care groups of the ARISE, PRoCESS, and PROMISE trials. The only RBC transfusion recommendation is that for adults, in the absence of extenuating circumstances such as myocardial ischemia, severe hypoxemia, or acute hemorrhage, RBC transfusion occurs only when hemoglobin concentration decreases to <7.0 g/dL in. The rationale for this recommendation following acknowledgment of the lack of difference between the outcomes of the EGDT and usual care arms of the above three studies states “We judge the evidence to be high certainty that there is little difference in mortality, and, if there is, that it would favor lower hemoglobin thresholds.”

Summary

Hematologic manifestations of sepsis are leukocytosis, leukopenia, lymphopenia, anemia, thrombocytopenia, and subclinical or clinical disseminated intravascular coagulation. Leukocytosis is expected but leukopenia may be seen and carries a worse prognosis. Lymphopenia when persisting in the septic patient may imply immunodeficiency and risk for additional infections, ongoing apoptosis, and worse outcomes. Thresholds for platelet transfusion are based on need for procedures and active bleeding. When sepsis-induced tissue hypoperfusion and lactic acidosis exist, red cell transfusion may be entertained as a method to increase oxygen delivery to the tissues as an alternate or adjunct to increasing cardiac output. The pendulum for use of blood transfusion as part of the initial resuscitation protocol of sepsis-induced tissue hypoperfusion, although still somewhat controversial, has swung away from transfusion unless the Hgb is less than 7 g/dL or there is evidence of cardiac ischemia or active hemorrhage. Subclinical DIC can be found in the majority of patients with severe sepsis, but clinical manifestations of DIC are uncommon.