Artificial mechanical red cells can give physicians the ability to precisely control saturation curve profiles independently for oxygen and carbon dioxide, either to maximize gas transport efficiency or to meet specialized demand functions imposed by emergency situations, unusual activities, or specific medical treatments.
4.1 Minimum Therapeutic Dose
The average male human body has 3 x 1013 red blood cells, each containing 2.7 x 108 hemoglobin molecules binding four O2 molecules per hemoglobin, giving a total human blood storage capacity of 3.24 x 1022 O2 molecules. Since hemoglobin normally operates between 95% saturation (arterial) and 70% saturation (venous) (Table 1), only 25% of stored O2 is accessible to the tissues, so active capacity of human blood is 8.1 x 1021 O2 molecules.
By contrast, each respirocyte stores 1.51 x 109 O2 molecules, 100% of which is accessible to the tissues, so full duplication of human blood active capacity requires deployment of 5.36 x 1012 devices. If respirocytes are administered via hypodermal injection or transfusion as a hematocrit-neutral 50% aqueous colloidal suspension, this implies a standard ~5.61 cm3 therapeutic dose of activated suspension, requiring only seconds to inject at, say, an accident scene. The therapeutic dose can duplicate natural red cell function indefinitely if the patient is breathing, and can supply all respiratory gas requirements from onboard storage alone for nearly 2 minutes for patients who are not breathing.
The alveolar membrane of human lungs can transmit a maximum of 3.2 x 1021 O2 molecules/sec across its surface , enough to fully load 42% of all therapeutic respirocytes during one transit time (5-10 sec ) of lung capillaries. Body blood circuit time is ~60 sec under resting conditions , so only 17% of injected respirocytes are present in the lungs at any one time. And since respirocytes can establish higher osmotic gradients than natural red blood cells, the rate of alveolar oxygen diffusion should not limit reoxygenation of an exhausted therapeutic dose.
Human O2 uptake  ranges from 0.48 kg/day at rest (whole-body heat generation ~100 watts, 2000 Kcal/day) to 9.6 kg/day at peak physical exertion (~1600 watts, 33,000 Kcal/day). Each exogenous 300 watts (6000 Kcal/day) added to the body's thermal load (e.g. by the operation of respirocytes) raises core temperature about 1 °F . A therapeutic dose of respirocytes releases from 17 watts (resting rate) to 340 watts (peak rate) while transporting respiratory gases, producing, at most, a 1 °F rise in core body temperature. Withdrawals from endogenous glucose stores to power continuous respirocyte activities amount to 350-7000 Kcal/day, ~20% normal caloric consumption, though this can probably be reduced to ~1% or less by employing a more energy-efficient design (Section 2.2.2). In the presence of normally functioning red blood cells, respirocytes remain dormant and their supplemental caloric input requirement is ~zero.
One of the potential benefits of nanomedical devices is their ability to extend natural human capabilities. If a physician wishes to permanently maximize the oxygen-carrying capacity of his patient's blood by infusing the largest possible number of respirocytes, the post-infusion equilibrium total hematocrit may be the principal limiting factor. The human male hematocrit has a normal range of 40%-52%, average 46% . Andean natives in high-altitude regions of Peru and Bolivia have adapted to thin air with hematocrits as high as 60% . Polycythemia patients present hematocrits of 75%-85% , leading to disturbances in blood flow as blood viscosity increases with rising hematocrit.
Infusion of 1 liter (~asymptomatic blood loss limit , ~plasma volume increase in high-altitude-adapted people , ~hemodilution limit for freshwater aspiration in the lungs , ~splenic capacity) of 50% respirocyte suspension raises total oxygen carrier volume to 55% of blood volume (RBC hct ~46% plus respirocrit ~9%), after absorption of water from the blood equal in volume to the infusion. This is probably near the maximum safe level for hematologically intact patients, as normal hematocrits rarely exceed ~52%.
In the absence of respiration or atmospheric oxygen, a fully-O2-charged augmentation dosage consisting of 9.54 x 1014 respirocytes could provide tissue oxygen delivery and carbon dioxide removal for 12 minutes at peak exertion and 3.8 hours at rest even during cardiostasis. (Dosage performance may be boosted to 1-20 hours by relaxing several conservative design constraints, such as raising storage pressure from 1000 atm to 10,000 atm and switching to unballasted devices with a single-chamber design.) The supplemental energy requirement to refill all respirocyte oxygen and glucose tanks after complete exhaustion is only 28 Kcal (~1 lump of sugar). Allowing at least 6 minutes for system refill holds heat generation below 300 watts (1 °F core temperature elevation; Section 4.1). Maximum alveolar oxygen transmission capacity may restrict recharge time to a minimum of 8 minutes. Refill requires absorption of 1.44 x 1024 O2 molecules (~53.7 liters O2 @ STP), available in inspired air in a time ranging from 6 minutes if the patient hyperventilates (62/min @ 3290 cm3 ) to 3 hours if respiration is shallow and normal (18/min @ 350 cm3).
The following infusion scenario appears plausible: After manufacture [2, 99], respirocytes comprising a single augmentation dosage are stored as a dry powder, tanks empty, in sealed plastic drip bags with two hose couplings. With no batteries to run down, consumables to age, vapors to outgas or organic matter to decompose, the product should have a long shelf life. To use the product, the bag is filled with ice-cold 0.13 M glucose (23 gm/liter) solution, plus any salts, minerals, vitamins, proteins or other substances the physician deems appropriate. The powder floats on the liquid surface. An external source of O2 gas (and CO2, if required) is provided through the second coupling. Respirocyte sensors detect the presence of glucose and begin pumping fuel into the glucose tanks. As these tanks fill, each device loads its oxygen tank to rated capacity. The powder still floats on the surface. Finally, the respirocytes load ballast tanks and sink to the bottom of the bag. (Powder remaining on the surface evidences malfunction.) During this ~30 sec charging process, the respirocytes absorb 53.7 liters of O2 @ STP, store 0.78 liters of CO2 waste, release enough energy to warm bag contents to 42 °C, and leave behind a 0.005 M glucose solution, closely matching normal blood concentration and temperature. Upon receiving the correct activation command, broadcast acoustically using an ultrasound transmitter device pressed against the bag, the respirocytes blow sufficient ballast water to achieve neutral buoyancy, creating a perfect suspension (after agitation) ready for IV drip. The suspension is ~300 times plasma viscosity (~castor oil or canola oil at 37 °C), still permitting ready plug flow . Similar warm-up procedures may be employed to administer smaller therapeutic doses, or small batches of sterilized precharged product may be kept in the office, ambulance, or hospital emergency room.
Larger doses are possible. Patients losing ~2 liters of blood may recover if volume is restored by a transfusion of blood or plasma ; recovery is also possible after loss of ~3 liters for gradual exsanguination over a period of 24 hours [97,101]. In emergency, shock, or battlefield situations, several liters of respirocyte suspension could be quickly transfused temporarily providing up to 12 hours of respiratory gas transport for resting nonbreathing patients. In theory, it would be possible to remove all red cells from circulation and permanently replace the entire 5.39-liter male human blood volume with respirocyte suspension. However, this procedure would be unwise for several reasons: (1) A 1-liter augmentation dose increases blood O2 storage capacity 18,000%, but "upgrading" to full replacement adds only another 400% while eliminating the natural red cell "backup" system, greatly increasing risk and violating the principles of fail-safe design and minimal intervention; (2) eliminating all red cells (which preferentially take up axial flow) may prevent platelet and leukocyte margination, interfering with natural hemostasis and immune functions (Section 5.6); (3) large systems of nanodevices may malfunction in unforeseen ways, increasing vulnerability to systematic design defects; and (4) anerythropoiesis (Section 5.2) may precipitate organ disturbances and biochemical imbalances such as (a) hemochromatosis and cumulative iron toxicity encouraging free-radical damage, leading to heart and liver failure, kidney failure, diabetes and hepatitis, (b) homeostatic responses to anemia including forceful/rapid heart action and pulse, vasodilation, volumetic change, etc., and (c) splenic disease.
Respirocyte behavior is initially governed by a set of default protocols which can be modified at any time by the attending physician. Besides the simple warmup procedure outlined in Section 4.2 and the filtration protocol described in Section 2.2.5, basic protocols will exist for operating molecular sorting rotors at various speeds and directions in response to sensor data. For example, ballast water pumping will normally be driven by internal ullage and temperature sensors. O2 rotors may load tanks at PO2 > 95 mmHg (alveolar) and unload at PO2 < 40 mmHg (tissues). CO2 rotors may fill at PCO2 > 46 mmHg and unfill at PCO2 < 40 mmHg, and may incorporate other sensor data including PO2, temperature, etc., to fine-tune pressure thresholds and enhance reliability. Gas loading parameters may be precisely specified in an individualized onboard lookup table provided by the physician for his patient, as for instance to adjust for declining arteriovenous oxygen gradient at high altitudes . Respirocytes, like natural hemoglobin, may also participate in the elimination of CO and in NO-mediated vascular control  if appropriate sorting rotors and onboard tankage are provided.
Respirocytes can be programmed with more sophisticated behaviors. Detection of PCO2 < 0.5 mmHg and PO2 > 150 mmHg, indicating direct exposure to atmosphere and a high probability that the device has been bled out of the body, should trigger a prompt gas venting and failsafe device shutdown procedure. Self-test algorithms monitoring tank filling rates, unaccounted pressure drops (indicating a leak), clutch responses, etc. may detect significant device malfunction, causing the respirocyte to place itself in standby mode ready to respond to an acoustic command to execute the filtration protocol for exfusion. Using onboard (21 nm)3 pressure transducers, respirocytes can detect, record, or respond to changes in heart rate or blood pressure, since arterial pressure is normally 0.1-0.2 atm (~10-19 joule/(21 nm)3 sensor) and systolic/diastolic differential is 0.05-0.07 atm (4-6 x 10-20 joule/(21 nm)3 sensor), both well above the mean thermal noise limit of kT ~ 4 x 10-21 joule at human body temperature . Outmessaging protocols could allow the population of respirocytes to communicate systemwide status (e.g. "low oxygen," "low glucose," "under immune attack," "cyanide detected" ) directly with the patient by inducing recognizable physiological cues (fever, shivering, gasping), or with the physician by generating subtle respiratory patterns requiring diagnostic equipment to detect, either automatically or in response to an acoustically-transmitted global inquiry initiated by patient or physician.
Other useful protocols may enable in vivo respirocytes to be commanded to cease or resume operating locally/globally, to enable/disable one or another class of sorting rotors, to alter sensor sensitivities or pressure actuation thresholds to control erythropoiesis (Section 5.2), to blow tanks or run all engines continuously, etc. Command messages of these types clearly must require lengthy authorization codes to prevent accidental or malicious triggering of respirocyte behaviors having potentially harmful consequences.
The safety and effectiveness of respirocyte devices will critically depend upon their mechanical reliability in the face of unusual environmental challenges, and upon their biocompatibility with human organs, tissues, and biochemical systems.
5.1 Mechanical Failure Modes
Basic self-diagnostic routines should be able to detect simple failure modes such as jammed rotor banks, plugged flues, gas leaks, and so forth, and to use backup systems to place the device into a fail-safe dormant mode pending removal by filtration. However, several potentially catastrophic physical failure modes deserve further discussion.
5.1.1 Device Overheating
Four scenarios involving device overheating include: (1) A normally-functioning respirocyte is deposited in a relatively dry (e.g., osteal or cartilaginous) location, losing contact with the aqueous heat sink; (2) one or more glucose engines become jammed at full open throttle; (3) inbound sorting rotors overload pressure vessel; and (4) device combustion.
In the first scenario, and assuming the thermal shutoff protocol fails, one glucose engine may continue to operate at full output for 10 seconds, emptying its fuel tank. This generates 6.1 picojoules, causing a maximum temperature rise in an isolated respirocyte ranging from +5.5 °K (full tanks) to +32.9 °K (empty tanks), insufficent to cause biological damage.
In the second scenario, and assuming failure of diagnostic protocols designed to detect and prevent multiengine runaway, the worst case occurs when all 12 glucose engines jam open at full throttle, generating 7.3 picowatts of continuous thermal energy. If this malfunction occurs while the respirocyte maintains contact with the aqueous heat sink, the device remains in thermal equilibrium at ~environmental temperature. (The process is self-limiting during capillary transits due to exhaustion of local free serum glucose supplies.) If the malfunction occurs when the respirocyte has been deposited in a thermally isolated location, device temperature rises until maximum operational limits are reached (perhaps ~400-500 °K ), moving parts fail and the device becomes inert. There is little danger of tank rupture because conservative pressure vessel design provides at least a 10,000% safety margin (Section 2.2.1) even at temperatures as high as 1300 °K .
In the third scenario, again assuming diagnostic protocol failure, inbound sorting rotors pump tanks to well above the 1000 atm maximum design pressure. This should not cause tank rupture, since sorting rotors can only pump to 30,000 atm  and the pressure vessel can withstand > 100,000 atm before bursting.
Finally, the complete combustion in air of one respirocyte (a full O2 chamber contributes less than 10% of the O2 required) releases at most 12,000 picojoules, barely enough energy to heat one tissue cell ~1 °C. Low maximum spark gap voltage across the minimum oxygen chamber gap precludes spontaneous device ignition. Accidental scalding, electrocution, or electromagnetic irradiation of respirocyte-bearing tissue should not trigger device combustion. And respirocytes can be rendered completely nonflammable by constructing the device internally of sapphire, a material with chemical and mechanical properties similar to diamond.
5.1.2 Noncombustive Device Explosion
Each respirocyte can contain up to 0.24 micron3 of O2 and CO2 gas at 1000 atm pressure, representing 24 picojoules of stored mechanical energy. If the device explodes in air the ejecta travel outward at a mean velocity of 257 m/sec, faster than the mean thermal velocity of 1-nm 310 °K particles (56 m/sec) but slower than the speed of sound in air (331 m/sec), so there should be no acoustic shock wave. If the device explodes inside human tissue, the gases do work against the surrounding fluid, displacing 10-16 m3 of water while raising its temperature 0.04 °K and expanding to a bubble 6 microns in diameter, just 2% the volume of a typical human cell. Since the average separation of neighboring respirocytes in the blood is 2-10 microns (depending on dosage), such gas bubbles should be reabsorbed almost immediately, so single-device explosions are unlikely to cause embolic or other significant damage.
Short of manufacturing defect, it is difficult to imagine a scenario that would lead to complete structural failure in vivo. Patients suffering a multistory fall onto a concrete pavement or a high-speed head-on automobile collision experience instantaneous accelerations of 102-104 g [104-105], but a spherical diamondoid shell should be able to resist accelerations of up to 108-1010 g. Crushing respirocyte-impregnated human tissue in an hydraulic press is unlikely to destroy any devices, as they will simply slide out of the way. The same logic applies to gunshot wounds, knife accidents that cut deep to hard bone, and blunt object blows to the skull .
Respirocytes will explode only if crushed between two hard planar surfaces with no avenue of escape. Skeletal joints roll in synovial fluid isolated from the bloodstream, so the only plausible scenario is dental grinding -- tooth enamel is the hardest natural substance in the human body, and a patient with an oral lesion could spread respirocyte-impregnated blood over the teeth. Single-device explosions may not be detectable; several thousand at once might produce a "fizziness" in the mouth. Simultaneously crushing 20 million respirocytes (the count in a 0.5-mm droplet of augmentation-dose blood) could produce a maximum jaw-speed (0.1 m/sec) explosive impulse, but this requires the out-of-bloodstream protocol to have failed simultaneously in all devices, an extremely unlikely event.
5.1.3 Radiation Damage
Simple analysis of nanodevice radiation damage and component reliability (background ~0.5 rad/year)  suggests that a respirocyte of active mass ~10-16 kg can expect 0.05 fatal radiation hits/year, implying an average operational lifespan of 20 years.
5.2 Interference with Erythropoiesis
The circulating red cell mass is homeostatically maintained by adjustments in the rate of red cell production (~1%/day of all red cells are destroyed naturally) in response to serum oxygenation. Hypoxic conditions in kidney peritubular cells  cause secretion of the hormone erythropoietin (EPO), inducing erythroid progenitor cells to differentiate into pronormoblasts, expanding the erythroid marrow, increasing red cell production and tissue oxygen levels [94,107].
The presence of a therapeutic dose of working respirocytes precludes tissue hypoxia, possibly reducing EPO secretion as low as 1% of normal levels , suppressing erythropoiesis. In a continuous antihypoxic environment, red cell hematocrit could fall to 14% , possibly ~0%. A similar mechanism, in which a shift to the right of the hemoglobin-O2 dissociation curve leads to increased tissue-O2 delivery, decreased erythropoietin output and declining hematocrit, has been proposed to explain sports anemia .
It may be possible to avoid decimation of the natural erythrocyte population by adjusting respirocyte PO2/PCO2 response thresholds so that these devices activate only when red cells are stressed. (There is additional risk of hyperoxic hemolysis [110-111] if device thresholds allow excessive O2 release.) Patients who have undergone large-volume respirocyte transfusions in emergency or battlefield situations may require posttraumatic recombinant EPO injections [112-113] to stimulate erythropoiesis, followed by exfusion of excess respirocytes as red cell hematocrit returns to normal.
An early hypothesis held that surface charge was the primary physicochemical feature of blood-contact material surfaces in determining thromboresistance. Cell coats with negatively charged sialic acid termini on both the glycoproteins and gangliosides, and macromolecules in all known flowing biological fluids, carry a slightly negative charge, thus should be repelled by bloodborne device surfaces bearing a net negative charge, or negative electrochemical potential, reducing the risk of thrombosis.
However, it is now known that immersion of adherent particles in liquid virtually eliminates electrostatic image forces, greatly reduces electrostatic contact potential forces, and can reduce van der Waals forces by at least a factor of six at organic-water interfaces . The hypothesis is further weakened when the immersion fluid is not neutral or insulating, but rather consists of the high ionic strength, salty, highly conductive ("dead short") biofluids actually found in vivo. Careful experiments designed to measure surface potential and surface charge found that the choice of surface electrical properties of materials intended to be brought into contact with blood or other salty aqueous fluids has little influence on biological adhesion . Indeed, there is growing evidence that any state of surface electrification is associated with greater rather than lesser accumulations of biological debris on such surfaces, though with an obvious accentuation of adhesive induction by net positive surfaces . Hence a net neutral respirocyte surface is preferred to minimize surface electrical thrombogenicity.
5.4 Mechanical Thrombogenesis
Spinning sorting rotors are unlikely to cause direct physical damage to formed blood elements, for several reasons. Rotors are atomically smooth and recessed into the housing, reducing physical contact with colliding surfaces and eliminating potential nucleation sites that may trigger blood clotting, gas embolus formation, or foaming. Only one pumping station at a time may have active spinning rotors (~0.5% of respirocyte surface), further reducing the likelihood of physical impact. Contact, when it occurs, should be relatively benign: Maximum rotor rim velocity is less than 1% of aortic blood velocity, and mean rotor velocity (incorporating dosage, workload and duty cycle) lies at or below capillary blood speed (Table 6).
Collisions between blood platelets and respirocytes are of similar velocity and frequency, and of much lower shear stress, as other collisions normally experienced by platelets with natural blood elements, hence respirocytes should not significantly increase the risk of mechanical thrombosis. Assuming geometrical cross-sections and thermal velocities, the most fragile elements, bloodstream platelets (~2 micron diameter ), should feel 7 hits/sec @ 5 mm/sec relative velocity by therapeutic respirocytes, a minor increment over the 2 hits/sec @ 1 mm/sec from other platelets, 40 hits/sec @ 0.4 mm/sec from red cells and ~100 hits/sec @ 0.5-5 mm/sec on capillary walls. Augmentation dosages will induce higher collision rates ~103 hits/sec @ 5 mm/sec, further enhanced because red cells disproportionally occupy blood vessel axial regions  (forcing platelets and respirocytes together toward the periphery) and because erythrocyte flip-flop motions  impart additional radial energy to platelets and respirocytes. However, the maximum shear stress per collision < 1 dyne/cm2. The threshold limit for shear stress-induced platelet aggregation is 60-90 dynes/cm2 [117-120], compared to a time-averaged 15-20 dynes/cm2 [117, 121] (range 5-56 dynes/cm2 ) for blood circulation in normal vessels, up to 100-400 dynes/cm2 reached when small arteries and arterioles are partially occluded as by atherosclerosis or vascular spasm [120, 122], so platelet aggregation, impact damage to circulating red cells or leukocytes, or vascular injury are unlikely. Results of vena cava, renal embolus, and ex vivo tests of ion-beam-deposited diamond-like carbon (DLC) coatings indicate a high degree of thromboresistance .
Could the extending cellular glycocalyx (not repelled by the neutral respirocyte surface) get trimmed, even by a recessed rotor? Rotor binding sites for glucose and small respiratory molecules of < 20 atoms involve pockets measuring < 2.7 nm in diameter , physically too small to accommodate the 10-20 nm thick cell coat layer or glycocalyx projections typically measuring 5-8 nm thick and 100-200 nm long , consisting of glycoproteins of 10,000 atoms or more. While an occasional sugar residue may get clipped, binding sites can be designed for maximum steric incompatibility with glycocalyx glycoproteins and proteoglycans, further minimizing the opportunities for trimming. Assuming 0.5% rotors active and ~0.1 msec collision duration, each bloodborne glycocalyx projection encounters an active rotor pocket only once every ~1-100 sec (usually without effect), depending upon dosage. As the cell coat is a secretion product incorporated into the cell surface that undergoes continuous renewal, any trimmed glycocalyx glycoproteins would be rapidly replaced via biosynthesis in the ribosomes of the endoplasmic reticulum, followed by final assembly with the oligosaccharide moiety in the Golgi complex.
5.6 Red Cell Aggregation and Non-RBC Margination
When aggregated, red cells preferentially take up axial flow and induce margination of platelets and leukocytes, increasing endothelial adhesion and other cell-wall interactions. Might respirocytes prevent aggregation or reduce margination, adversely affecting thrombotic and other natural defense mechanisms?
Normal wall shear rates in physiological bloodflow range from 50-700/sec in the larger arteries to 250-2000/sec in the smaller arteries and capillaries . At shear rates < 50/sec , red cells form linear and branched chain aggregates (rouleaux) in which cell surfaces are cross-linked by fibrinogen. Since respirocytes should exhibit no axial preference , and since their diamondoid surfaces are not biochemically active and possibly can be rendered nearly invisible to fibrinogen (Section 5.7), their presence should not interfere with adhesive processes involved in axial rouleaux formation far from vessel walls, where shear rates are lower.
As red cells aggregate in tube flow, rouleaux migrate inward forming a network of aggregates in the core of the tube surrounded by a peripheral cell-depleted layer consisting of single cells, occasional small rouleaux, white cells, platelets, and respirocytes. This results in a two-phase flow of a relatively high shear rate peripheral zone surrounding a low shear rate, high cell concentration, central zone . Therapeutic dose respirocytes represent ~0.1% of total red cell mass, up to only ~22% for 1-liter augmentation doses, so collisional dispersion of the high-shear peripheral zone should be minimal and effective diffusion rates should remain high. Since a plasma suspension of microspheres at a 9% particle volume fraction (~final respirocrit of maximum augmentation dose) has a viscosity indistinguishable from 9% hematocrit blood , and since 55% hct blood does not disrupt margination, respirocytes should pose no danger in this regard.
5.7 Coagulation, Inflammation and Phagocytosis
The initial event which dictates biocompatibility upon implantation of a foreign material is adsorption of plasma proteins from blood onto the surface [126-128]. Since protein adsorption is much more rapid than the transport of cells to foreign surfaces, host cells interact with host proteins adsorbed on the material surface rather than with the foreign material itself. The adsorption and conformational state of fibrinogen, the major surface protein to initiate coagulation [126,129-131] and inflammation [132-133], is commonly used as a biocompatibility indicator .
The biocompatibility of diamond surfaces has not been extensively investigated. A recent study  of chemical-vapor-deposited (CVD) diamond surface adsorption of elutable human fibrinogen showed sample adsorption rates of ~200 nanograms/cm2, equivalent to ~104 molecules coating a 1-micron sphere, adding ~1% to respirocyte mass and coating ~10% of device surface. The high surface energy of diamond makes it extremely hydrophobic, so fibrinogen (and formed blood elements) should not readily adhere to atomically-precise respirocyte diamond surfaces [123,136] (e.g. ocean barnacles do not attach to diamond ). If plasma protein adsorption remains a problem, each respirocyte can be equipped with a fully active surface able to mechanically desorb fibrinogen (molecule covers ~1 sorting rotor area at attachment point), or the surface can be heparinized [138-139], coated with synthetic glycolipids such as polyethyleneglycol , or covalently bound with masking groups (such as coat components of young erythrocytes which are invisible to the reticuloendothelial system (RES) ) or macrophagophobic molecules. Complement activation and stimulation of febrile or allergic reactions by diamond have yet to be tested.
The rate of attachment of human polymorphonuclear leukocytes (the most abundant white cell in human blood) to CVD diamond surface ~400,000 cells/cm2, or ~1 cell per 100 respirocytes, even after pre-coating with human plasma for 4 hours . SEM photographs of CVD diamond surfaces implanted intraperitoneally in mice for 1 week showed no inflammatory response; polished surface (~1000-nm features) showed some spread and fused macrophages, but unpolished surface (~250-nm features) showed round, non-spread (non-activated) macrophages without psuedopodia and cell bridges . It is possible that atomically-perfect diamond surfaces (~1-nm features) are nearly invisible to macrophages. Another study using tissue culture plates with 400-nm-thick DLC coatings elicited no toxic or inflammatory response in mouse peritoneal macrophages and mouse fibroblasts grown on their surface . In vivo experiments implanting DLC-coated orthopedic pins in sheep also showed low bioactivity . Further research is needed to confirm whether flawless diamond surfaces may be made entirely bioinactive.
Early experiments with artificial red blood cells showed that cells larger than 1 micron were removed from the pulmonary circulation, while those 1 micron or smaller were removed by the RES , particularly Kupffer cells in the liver. If Kupffer cells could recognize respirocytes and attempted to phagocytose them, as little as ~0.4 gm (14% of a therapeutic dose) of these non-metabolizable particulates might be sufficient to blockade the mononuclear phagocytic system and halt the process . But large numbers of indigestible respirocytes trapped in the liver and spleen could lead to hepatomegaly, splenomegaly, or organ necrosis, so intentional nonspecific blockade should be avoided in respirocyte design. Functional devices prematurely ingested by macrophages may execute protocols designed to free themselves , such as emitting gases (O2 oxidizes and inactivates many cell components such as essential sulfydryl groups of enzymes; CO2 toxifies the cell via acidosis), absorbing cellular O2, H2O or glucose (asphyxiating, dehydrating or starving the captor), or purging gas tanks (blowing a 6-micron bubble, ~25% of the volume of a ~10 micron neutrophil, inducing "indigestion").
However, particulate matter not coated with blood proteins does not adhere to the 70-nm fuzzy coat lining the cells of hepatic sinusoids . Under normal conditions, formed elements of the blood, lipid droplets like chylomicrons, etc. also do not adhere to the wall of sinusoids . If diamondoid surfaces can be rendered unwetted or unwettable by fibrinogen and other blood proteins, or can be appropriately masked, then physically intact respirocytes should not be recognized (or phagocytosed) by Kupffer cells or macrophages, nor induce giant cell formation. Hepatic and splenic macrophages digest dysfunctional red blood cells but probably cannot recognize rapidly spinning synthetic sorting rotor binding sites, which are deeply embedded in passive diamondoid structure.
Intact but bioinactive respirocytes smaller than red cells should readily pass both spleen (where blood percolated through a macrophage bed in the red pulp reenters the bloodstream through ~3 micron slits) and liver. Respirocytes are much too large to diffuse through the capillaries of the renal glomeruli and 60-nm renal endothelial wall openings, hence should pass the kidney unfiltered. Intact devices should remain confined largely to the bloodstream so few will enter the lymphatic system, except possibly via the lungs.
Respirocytes that have lost physical integrity should be recognized as foreign matter (and be engulfed by free macrophages or Kupffer cells) because their rough exposed surfaces will invite blood protein tagging. Fragment removal from blood takes place almost exclusively in the liver and red pulp of the spleen. Metal and dye particles are not metabolized and remain in the RES almost indefinitely . Respirocyte fragments, for the most part, will consist of sharp indigestible shards that quickly destroy the ingesting macrophage, causing it to rupture and discharge its cytoplasmic contents. This will lead to acute local inflammation and probable release of chemotactic agents attracting mesenchymal cells to the site, which will then differentiate into fibroblasts, resulting in entombment of the shards in the adjacent tissue by a permanent fibrous spherical granulomatous capsule . Fortunately, disintegration of respirocytes should be an exceedingly rare event (Section 5.1).