CROSS-REFERENCE TO RELATED APPLICATIONSThis application is a continuation of U.S. application Ser. No. 11/157,549, filed on Jun. 21, 2005, which is a continuation-in-part from U.S. application Ser. No. 11/050,611, filed on Feb. 3, 2005, which claims priority from U.S. Provisional Application No. 60/541,004, filed on Feb. 5, 2004.
This application is related to the application entitled “Self-Improving Classification System” and “Self-Improving Identification Method,” which were filed on Jun. 12, 2005 and also assigned to Medtronic, Inc.
BACKGROUND OF THE INVENTIONThe present invention relates to a system and method for identifying candidates for receiving cardiac therapy based on biochemical markers associated with propensity for arrhythmias.
Many patients experiencing ventricular tachyarrhythmia may be at risk of loss of heart function. Sudden cardiac death, which results from a loss of heart function, is often preceded by episodes of ventricular tachyarrhythmia such as ventricular fibrillation (VF) or ventricular tachycardia (VT). Many patients are unaware that they are at risk of ventricular tachyarrhythmia. For some unfortunate patients, a sudden cardiac death incident may be the first sign that they were at risk. It is, of course, preferable for such patients to be aware of their risk in advance of such an event. In patients who are aware of their risk, an implantable medical device, such as a pacemaker with defibrillation and cardioverson capability, may drastically increase the survival rates of such patients.
BRIEF SUMMARY OF THE INVENTIONIn general, the invention is directed to systems and techniques for assessing a risk of ventricular tachyarrhythmia in a patient. In some medical conditions, including but not limited to ventricular tachyarrhythmia, certain biochemical factors in the body of the patient reflect the health of a patient. A patient that experiences ventricular tachyarrhythmia, for example, may experience an increased or decreased concentration of identifiable proteins in his/her blood, even if the patient is symptom free. By measuring the concentration of these biochemical markers or “biomarkers” in the patient, an assessment of a risk of ventricular tachyarrhythmia for the patient can be made, based upon the measurements.
BRIEF DESCRIPTION OF THE DRAWINGSFIG. 1 is a conceptual logical diagram illustrating an embodiment of the invention.
FIG. 2 is a conceptual logical diagram illustrating a variation of the embodiment of the invention shown inFIG. 1.
FIGS. 3 and 4 are flow diagrams illustrating techniques for assessment of risk of ventricular tachyarrhythmia.
FIG. 5 is a conceptual diagram illustrating a technique for mass spectral analysis of a sample for biochemical markers.
FIG. 6 is a graph showing differences in biochemical marker abundance for a patient at risk of ventricular tachyarrhythmia, compared to a patient in a control group.
FIG. 7 is a logical diagram illustrating a technique for sorting patients at risk of ventricular tachyarrhythmia from a control group.
FIG. 8 is a logical diagram illustrating a technique for classifying patients at risk of ventricular tachyarrhythmia.
FIG. 9 is a block diagram of a system configured to carry out an embodiment of the invention.
DETAILED DESCRIPTION OF THE INVENTIONFIG. 1 is a conceptual logical diagram illustrating an embodiment of the invention. Based upon measuring one or more biochemical markers in a group ofpatients 10, the invention provides for assessing a risk of ventricular tachyarrhythmia in each patient as a function of the measurement.
In the illustration shown inFIG. 1, a “tree analysis” sorts the patients into groups according to measurements of three biochemical markers. The biochemical markers are identified by the letters “A,” “B,” “C” and “D.” Typical biochemical markers include proteins (that includes, for example, peptides, polypeptides, and polyamino acids of any length or conformation), lipids, genes and peptides or any combination thereof, but the illustration shown inFIG. 1 is not limited to any particular biochemical marker or set of biochemical markers. Specific examples of biochemical markers are discussed below.
For each patient, a measure of a first biochemical marker (denoted MA) is determined. Determining the measure of biochemical marker “A” for a particular patient may include, for example, determining the concentration or mass of biochemical marker “A” in a standard sample of bodily fluid taken from that patient. For each patient, the measure of the first biochemical marker is compared to a threshold value (denoted TA). Those patients for whom MA is greater than or equal to TAare deemed to be in agroup12 that is not at significant risk of ventricular tachyarrhythmia, and no further testing need be done for the members ofgroup12. Those patients for whom MAis less than TAare deemed to be in agroup14 that may be, or may not be, at risk of ventricular tachyarrhythmia. InFIG. 1, the members ofgroup14 undergo further testing to determine the individual members' risks of ventricular tachyarrhythmia.
For each patient ingroup14, a measure of a second biochemical marker “B” (denoted MB) is determined. For each patient ingroup14, the measure of the second biochemical marker is compared to a second threshold value (denoted TB). Those patients for whom MBis less than TBare deemed to be agroup16 that is not at significant risk of ventricular tachyarrhythmia, and no further testing need be done for the members ofgroup16. Those patients for whom MBis greater than or equal to TBare deemed to be in agroup18 that may be, or may not be, at risk of ventricular tachyarrhythmia.
The members ofgroup18 undergo further testing with respect to a measure of a third biochemical marker “C” (denoted MC). For each patient ingroup18, the measure of the third biochemical marker is compared to a third threshold value (denoted TC). On the basis of the comparison, the patients are divided into agroup20 that is not at significant risk of ventricular tachyarrhythmia and agroup22 that is at significant risk of ventricular tachyarrhythmia.
In other words,FIG. 1 illustrates assessing a risk of ventricular tachyarrhythmia for a patient as a function of the measurement of three biochemical markers. Unless a patient meets the threshold criteria for all three biochemical markers, the patient will not be deemed to be at significant risk of ventricular tachyarrhythmia.
The thresholds TA, TBand TCare determined empirically. Clinical studies and experience may be used to determine thresholds for each biochemical marker. The thresholds may differ from marker to marker. For some biochemical markers, a patient may be at higher risk when the measure of the biochemical marker is above the threshold, and for other biochemical markers, the patient may be at higher risk when the measure of the biochemical marker is below the threshold.
FIG. 2 is a conceptual logical diagram illustrating an embodiment of the invention that is a variation of the technique illustrated inFIG. 1. UnlikeFIG. 1, patients sorted intogroup12 are subjected to further testing. For each patient ingroup12, a measure of a fourth biochemical marker “D” (denoted MD) is determined, and the measure is compared to a fourth threshold value (denoted TD). On the basis of this comparison, patients ingroup12 are sorted intogroups24 and26. Those patients ingroup24 are deemed to be not at significant risk of ventricular tachyarrhythmia, and no further testing need be done for the members ofgroup24.
Those patients ingroup26, however, are subjected to further testing. The members ofgroup26 undergo further testing with respect to the third biochemical marker “C,” just like the members ofgroup18. On the basis of a comparison of the measure of the third biochemical marker to the third threshold, the patients ingroup26 are divided into agroup28 that is not at significant risk of ventricular tachyarrhythmia and agroup30 that is at significant risk of ventricular tachyarrhythmia.
In other words,FIG. 2 illustrates assessing a risk of ventricular tachyarrhythmia for a patient as a function of the measurement of four biochemical markers. A patient may be deemed to be at significant risk of ventricular tachyarrhythmia according to more than one testing path.
FIG. 3 is a flow diagram illustrating logical sorting embodiments such as are depicted inFIGS. 1 and 2. An apparatus, such as an apparatus listed inFIGS. 5 and 6, or a technician measures a first biological marker (40) and assesses a risk of ventricular tachyarrhythmia in the patient as a function of the measurement (42). The apparatus or technician measures a second biological marker (44) and assesses the risk of ventricular tachyarrhythmia in the patient as a function of that measurement (46).
In the procedure outlined inFIG. 4, the apparatus or technician measures a first biological marker (50) and measures a second biological marker (52), and assesses the risk of ventricular tachyarrhythmia in the patient as a function of both measurements (54). The techniques shown inFIGS. 3 and 4 may achieve the same result; that is, a patient may be sorted according to risk of ventricular tachyarrhythmia using either technique. When a patient is deemed to be at risk, an appropriate therapy may be applied. Therapy for a patient may include, for example, implanting an electronic cardiac stimulation device in the patient that detects and terminates episodes of ventricular tachyarrhythmia or administering an antiarrhythmic drug that prevents induction of such episodes.
FIG. 5 is a conceptual diagram illustrating a technique for measuring a plurality of biological markers. Abiochip60 comprises asubstrate62 and one ormore elements64A. InFIG. 5, four distinct sensing elements are coupled tosubstrate62, but the invention encompasses use of any number of sensing elements.
Biochip60 is a set of miniaturized test sites, or microarrays, arranged on asolid substrate62 made from a material such as silicone or glass. Each test site includes a set ofsensing elements64A. In general, sensing elements include one or more components that change conformation in the presence of an analyte of interest. Typical sensing elements include antibody molecules that change conformation in the presence of a specific biomarker but that do not change conformation in the presence of any other biomarker. The invention encompasses any sensing element, however, and is not restricted to antibodies. The sensing elements ofbiochip60 may have general properties such as high affinity toward hydrophilic or hydrophobic molecules, or anionic or cationic proteins, for example.
Substrate62 may have a surface area of about one square centimeter, but the invention encompasses biochips that are larger or smaller.Substrate62 may be formed in any shape, may include any number of test sites, and may include any combination of sensing elements. The invention is not limited to any particular biochip.
Biochip60 is exposed tosample66.Sample66 may include any biological sample from a patient, such as a blood sample. Biomarkers present insample66 react with sensing elements onbiochip60.Exposed sensing elements64B typically react with biomarkers insample66 by undergoing a conformational change, or by forming ionic, covalent or hydrogen bonds. The unreacted or unbound portion ofsample68 is washed away.
The concentrations of biomarkers insample66 are a function of the extent of the reaction between exposed sensing elements64 andsample66. The extent of the reaction is determinable by, for example, mass spectrometry. The Surface Enhanced Laser Desorption/Ionization (SELDI) process is an example of a mass spectrometry technique for determining the concentrations of biomarkers that does not necessarily need antibodies. Instead, the molecules are absorbed onto a surface, and later released from the same surface with its energy absorbing matrix upon the application of external energy, usually in the form of light.
In general, the SELDI process directs light generated by one or morelight sources70 atbiochip60. Amass analyzer72 measures the molecular weight of the biomarkers. In particular, biomarkers onbiochip60 are ionized and separated, and molecular ions are measured according to their mass-to-charge ratio (m/z). Ions are generated in the ionization source by inducing either the loss or the gain of a charge (e.g. electron ejection, protonation, or deprotonation). Once the ions are formed in the gas phase, they can be electrostatically directed intomass analyzer72, separated according to their mass, and finally detected.
Proteins bound to sensingelements64B, for example, can be ionized and separated based on molecular properties, such as being hydrophilic versus hydrophobic. Proteins captured by sensingelements64B are freed by the energy provided by a weak laser pulse, and charged positively by the removal of a second electron as a result of illumination by a second laser pulse. Time of flight though a vacuum tube following acceleration in an electric field allows the measurement of the mass-to-charge ratio.
The invention supports other techniques for determining the concentrations of biomarkers and is not limited to the SELDI process. In one embodiment, for example, the techniques of the invention could be carried out by using conventional assays for individual biomarkers, such as an Enzyme Linked ImmunoSorbent Assay (ELISA tests). An advantage of using a biochip is that a biochip saves time and effort in comparison to individual assays when multiple markers are to be measured.
Many protein markers are generally accepted as being indicative of cardiac conditions. C-Reactive Protein (CRP) is associated with sudden cardiac death; Fatty Acid Binding Protein is a plasma marker associated with acute myocardial infarction; Cardiac Troponin is associated with myocardial infarction; Myosin Light and Heavy Chains are associated with heart failure; brain natriuretic peptide (BNP) is associated with left ventricular heart failure, and so on.
Other markers may be associated with other cardiac conditions of interest. The markers may be identified by their name, or by other characteristics, such as molecular weight.
In an example clinical study, patients with coronary artery disease were divided into two groups: a test group that had coronary artery disease, and an implantable medical device (IMD) (with one sustained VT/VF episode with cycle length less than or equal to 400 ms); and a control group having coronary artery disease but no IMD, and no known history of VT/VF. In the study, sixteen patients had an IMD and thirty-two were in the control group. Certain patients were excluded from the study, including non-Caucasians, females, patients outside of an age limit of 45-80, and patients having certain health problems or cardiac conditions. Patients meeting the inclusion criteria were enrolled in the study. Upon enrollment, an extensive questionnaire, including medical history, was filled.
Three blood samples were drawn from each patient. At least one sample comprised 8.5 mL of blood drawn from the patients for serum separation. Serum is the cell free portion of the blood containing proteins and lipids. At least one other sample of an additional 12 mL of blood was drawn and kept as whole blood for eventual genetic analysis. The samples were analyzed using proteomic and lipidomic techniques.
During processing, proteins in the serum were fractionated into four distinct groups based on pH (acidity) of the protein. Later on, these proteins were spotted onto three surfaces of one or more biochips. The surfaces had different chemical affinities. A surface designated “CM10” was responsive to weak cation exchange surface. A surface designated “H50” was a hydrophobic surface. A surface designated “IMAC” was an immobilized metal affinity surface. The SELDI time-of-flight technique was used to measure the molecular weight of the protein on each surface.
FIG. 6 shows the results of sample proteomic spectra of two patients, one having an IMD (80) and one in the control (82). These results indicate that some of the protein markers in the blood were expressed differently in two groups. Data produced by processing of all of the patients followed similar patterns, i.e., the data indicated that some of the protein markers in the blood obtained from patients were expressed differently in two groups. The differences in markers may form a basis for distinguishing the patients that would benefit from an IMD from the patients that would not benefit.
FIG. 7 shows a tree analysis applied to these results to identify potential biomarkers that differentiate patients who have a higher propensity for fatal ventricular arrhythmias from the others. As a result of the tree analysis, four protein markers could be used to classify the 48 patients correctly. Specifics of these protein markers are shown in the table below:
|
Protein | Molecular | Isoelectric pH | |
Number | Weight (Da) | (pl) | Capture Surface |
|
|
P1 | 10,146.5 | 9+ | CM10 weak cation |
| | | exchange) |
P2 | 15,006 | 9+ | CM10 weak cation |
| | | exchange) |
P3 | 166,582 | 5-7 | CM10 weak cation |
| | | exchange) |
P4 | 10,948 | 9+ | IMAC (Immobilized Ion |
| | | Affinity Surface) |
|
In the above table, proteins are identified by a number and are characterized by a molecular weight in Daltons and an Isoelectric pH (pI). The molecular weight in Daltons is not necessarily unique to any particular protein, but proteins are often distinguishable by molecular weight. It is not necessary to the invention that the protein having that molecular weight and/or pI be specifically identified by name or by amino-acid sequence.
As shown inFIG. 7, the amount of protein P1 in the serum was tested for allpatients90.Patients92 having an abundance of P1 greater than or equal to 1.0422237 (measured in arbitrary units) were not at significant risk of ventricular tachyarrhythmia and were, therefore, not candidates for an IMD.Patients94 having an abundance of P1 less than 1.0422237, however, could not be classified by abundance of P1 alone.
Forpatients94, the amount of P2 in the serum was tested.Patients96 having an abundance of P2 less than 0.2306074 were not candidates for an IMD.Patients98 having an abundance of P2 greater than or equal to 0.2306074 were tested for protein P3.Patients100 having an abundance of P3 greater than or equal to 0.0491938 were not candidates for an IMD, whilepatients102 having an abundance of P3 less than 0.0491938 were tested for protein P4.Patients104 having an abundance of P4 greater than 0.027011 were considered to be candidates for an IMD, while the remainingpatients106 were not considered to be candidates for an IMD.
The arbitrary units may be normalized to an abundant protein, such as albumin, which is generally consistent in relative abundance among a group of patients or by spiking the original sample with a known concentration of an exogenous substance, and scaling the entire spectrum such that the measured value of the exogenous compound matches the amount that was added to the sample. The invention supports the use of other bench marks as well, such as the total ion current in the mass spectrometer used to measure the protein abundance.
In addition, the invention supports a range of measurement standards. In some cases, it is not feasible to perform measurements that have one hundred percent sensitivity and specificity, and some standards may be applied to determine whether the patient is at significant risk of ventricular tachyarrhythmia or not. The tree analysis depicted inFIG. 7, for example, is generally more sensitive and specific than conventional patient sorting techniques (such as averaged electrocardiogram), even though it may result in some false positives and false negatives.
The tree shown inFIG. 7 may be generated using Classification and Regress Tree (CART) analysis. The tree analysis depicted inFIG. 7 is an example of an approach for assessing a risk of ventricular tachyarrhythmia in one or more patients as a function of the measurement of one or more biochemical markers. The assessment may be preformed in other ways as well. The test may be expressed as a logical test such as an IF-THEN test, which can be implemented in software:
| |
| IF |
| ((P1>1.0422237) AND (P2≧0.2306074) AND (P3<0.0491928) |
| AND (P4≧0.027011)) |
| THEN |
| PATIENT IS AN IMD CANDIDATE |
| |
This IF-THEN test gave the following results when applied to the clinical data where two samples from each patient were processed:
Sensitivity: 27/(27+5)=84%
Specificity: 63/(63+1)=98%
False Positives: 1/(1+27)=4%
False Negatives: 5/(5+63)=7%
Using conventional sorting techniques, sensitivity and specificity tend to be around 55 to 75 percent. This clinical data demonstrates an improvement in sensitivity and specificity in comparison to conventional techniques.
Another technique for assessing a risk of ventricular tachyarrhythmia in one or more patients as a function of a measurement of one or more biochemical markers is to use an artificial neural network. In an exemplary application, the clinical data was analyzed using an artificial neural network having four input nodes corresponding to proteins P1, P2, P3 and P4. The network included four hidden nodes and one output. This artificial neural network gave the following results when applied to the clinical data where two samples from each patient were processed:
Sensitivity: 24/(24+8)=75%
Specificity: 63/(63+1)=98%
False Positives: 1/(1+25)=4%
False Negatives: 8/(8+63)=11%
A second representative clinical study to discover class identifiers was carried out with an additional 30 patients. These additional patients also had coronary artery disease and met specific inclusion criteria. They were divided into the two groups based on whether or not patients had an IMD. The patients having an IMD also had at least one true VT/VF episode with a cycle length less than or equal to 400 ms terminated in the last 90 days. A total of 29 patients were in the test group, which consists of patients with an IMD, and 49 patients were in the control group.
Patients filled out an extensive questionnaire that included medical information that was then used in creating specimen/patient profiles. The following table shows the patient characteristics included the specimen profile and the relative breakdown between the test and control groups.
|
| Patients in ICD | Patients in | Total |
| Arm | Control Arm | Patients |
Patient Characteristics | (N = 29) | (N = 49) | (N = 78) |
|
Gender (N, %) | | | |
Male | 29 (100%) | 49 (100%) | 78 (100%) |
Female | 0 (0%) | 0 (0%) | 0 (0%) |
Age (years) |
Mean | 68.8 | 67.1 | 67.8 |
Standard Deviation | 8.2 | 8.1 | 8.1 |
Minimum-Maximum | 50-81 | 51-81 | 50-81 |
Left Ventricular |
Ejection Fraction |
Time since most recent |
LVEF (days) |
Mean | 1.1 | 0.9 | 1 |
Standard Deviation | 1.1 | 1.1 | 1.1 |
Minimum-Maximum | 0-5.3 | 0-4.8 | 0-53 |
Most Recent Documented |
Measurement (%) |
Mean | 37.9 | 51.2 | 46.2 |
Standard Deviation | 9.6 | 9.5 | 11.5 |
Minimum-Maximum | 28-66 | 29-73 | 28-73 |
Method of LVEF |
Measurement |
Radionucleotide | 6 (21%) | 19 (39%) | 25 (32%) |
angiocardiography/MUGA |
Echo | 9 (31%) | 16 (33%) | 25 (32%) |
Cath | 13 (45%) | 14 (29%) | 27 (35%) |
Unknown | 0 (0%) | 0 (0%) | 0 (0%) |
|
The next table shows patient cardiovascular surgical and medical history that was included in the specimen profiles and the relative breakdown between the control and test groups.
|
| Patients in | Patients in | Total |
| ICD Arm | Control Arm | Patients |
Patient Characteristics | (N = 29) | (N = 49) | (N = 78) |
|
|
Cardiovascular Surgical | | | | | | |
History (N, %) |
None | 6 | (20.7%) | 4 | (8.2%) | 10 | (12.8%) |
Coronary Artery Bypass Graft | 14 | (48.3%) | 20 | (40.8%) | 34 | (43.6%) |
Coronary Artery Intervention | 13 | (44.8%) | 36 | (73.5%) | 49 | (62.8%) |
Angioplasty | 8 | (27.6%) | 25 | (51%) | 33 | (42.3%) |
Stent | 7 | (24.1%) | 31 | (63.3%) | 38 | (48.7%) |
Atherectomy | 0 | (0%) | 0 | (0%) | 0 | (0%) |
Ablation | 3 | (10.3%) | 3 | (6.1%) | 6 | (7.7%) |
Valvular Surgery | 1 | (3.4%) | 1 | (2%) | 2 | (2.6%) |
Other | 1 | (3.4%) | 2 | (4.1%) | 3 | (3.8%) |
Cardiovascular Medical |
History |
None | 0 | (0%) | 0 | (0%) | 0 | (0%) |
Coronary Artery Disease | 29 | (100%) | 49 | (100%) | 78 | (100%) |
Myocardial Infarction | 29 | (100%) | 49 | (100%) | (78 | (100%) |
Number of infarctions | | | |
Mean | 1.4 | 1.3 | 1.3 |
Standard Deviation | 0.6 | 0.5 | 0.5 |
Minimum-Maximum | 1-3 | 1-3 | 1-3 |
Time Since First Infarction |
(years) |
Mean | 10.2 | 6.6 | 7.9 |
Standard Deviation | 7.7 | 5.6 | 6.6 |
Minimum-Maximum | 1-26 | 0-22 | 0-26 |
Time Since Most Recent |
Infarction (years) |
Mean | 5.3 | 5 | 5.1 |
Standard Deviation | 5 | 5.4 | 5.2 |
Minimum-Maximum | 1-17 | 0-22 | 0-22 |
Hypertension | 19 | (65.5%) | 34 | (69.4%) | 53 | (67.9%) |
Cardiomyopathy | 18 | (62.1%) | 3 | (6.1%) | 21 | (26.9%) |
Hypertrophic | 5 | (17.2%) | 0 | (0%) | 5 | (6.4%) |
Dilated | 9 | (31%) | 3 | (6.1%) | 12 | (15.4%) |
Valve Disease/Disorder | 4 | (13.8%) | 8 | (16.3%) | 12 | (15.4%) |
Aortic | 0 | (0%) | 5 | (10.2%) | 5 | (6.4%) |
Tricuspid | 1 | (3.4%) | 1 | (2%) | 2 | 92.6%) |
Mitral | 4 | (13.8%) | 3 | (6.1%) | 7 | (9%) |
Pulmonary | 0 | (0%) | 0 | (0%) | 0 | (0%) |
Primary/Idiopathic Electrical | 0 | (0%) | 1 | (2%) | 1 | (1.3%) |
Conduction Disease |
Documented Accessory | 0 | (0%) | 0 | (0%) | 0 | (0%) |
Pathway |
Chronotropic Incompetence | 1 | (3.4%) | 0 | (0%) | 1 | (1.3%) |
NYHA Classification |
Class I | 4 | (13.8%) | 3 | (6.1%) | 7 | (9%) |
Class II | 6 | (20.7%) | 5 | (10.2%) | 11 | (14.1%) |
Class III | 4 | (13.8%) | 0 | (0%) | 4 | (5.1%) |
Class IV | 0 | (0%) | 0 | (0%) | 0 | (0%) |
Not Classified | 15 | (51.7%) | 41 | (83.7%) | 56 | (71.8%) |
Congenital Heart Disease | 0 | (0%) | 0 | (0%) | 0 | (0%) |
Other | 2 | (6.9%) | 3 | (6.1%) | 5 | (6.4%) |
|
The table that follows shows patient arrhythmia history that was included in the specimen profiles and the relative breakdown between the control and test groups.
|
| Patients in | Patients in | Total |
| ICD Arm | Control Arm | Patients |
Patient Characteristics | (N = 29) | (N = 49) | (N = 78) |
|
|
Spontaneous Arrhythmia | | | | | | |
History (N, %) |
None | 0 | (0%) | 26 | (53.1%) | 26 | (33.3%) |
Ventricular |
Sustained Monomorphic | 23 | (79.3%) | 0 | (0%) | 23 | (29.5%) |
VT |
Sustained Polymorphic VT | 1 | (3.4%) | 0 | (0%) | 1 | (1.3%) |
Nonsustained VT | 17 | (58.6%) | 0 | (0%) | 17 | (21.8%) |
Ventricular Flutter | 1 | (3.4%) | 0 | (0%) | 1 | (1.3%) |
Ventricular Fibrillation | 9 | (31%) | 0 | (0%) | 9 | (11.5%) |
Torsades de Pointes | 0 | (0%) | 0 | (0%) | 0 | (0%) |
Long Q/T Syndrome | 0 | (0%) | 0 | (0%) | 0 | (0%) |
Other | 0 | (0%) | 2 | (4.1%) | 2 | (2.6%) |
Bradyarrhythmias/ |
Conduction |
Disturbances |
Sinus Bradycardia | 5 | (17.2%) | 14 | (28.6%) | 19 | (24.4%) |
Sick Sinus Syndrome | 0 | (0%) | 2 | (4.1%) | 2 | (2.6%) |
1° AV Block | 7 | (24.1%) | 6 | (12.2%) | 13 | (16.7%) |
2° AV Block | 0 | (0%) | 0 | (0%) | 0 | (0%) |
Type I (Mobitz) | 0 | (0%) | 0 | (0%) | 0 | (0%) |
Type II (Wenckebach) | 0 | (0%) | 0 | (0%) | 0 | (0%) |
3° AV Block | 0 | (0%) | 0 | (0%) | 0 | (0%) |
Right Bundle Branch Block | 5 | (17.2%) | 8 | (16.3%) | 13 | (16.7%) |
Left Bundle Branch Block | 4 | (13.8%) | 2 | (4.1%) | 6 | (7.7%) |
Bradycardia-Tachycardia | 0 | (0%) | 0 | (0%) | 0 | (0%) |
Syndrome |
Other | 2 | (6.9%) | 0 | (0%) | 2 | (2.6%) |
Atrial Arrhythmia History |
(N, %) |
None | 14 | (48.3%) | 36 | (73.5%) | 50 | (64.1%) |
Atrial Tachycardia | 4 | (13.8%) | 0 | (0%) | 4 | (5.1%) |
Paroxysmal | 4 | (13.8%) | 0 | (0%) | 4 | (5.1%) |
Recurrent | 0 | (0%) | 0 | (0%) | 0 | (0%) |
Chronic | 0 | (0%) | 0 | (0%) | 0 | (0%) |
Atrial Flutter | 1 | (3.4%) | 5 | (10.2%) | 6 | (7.7%) |
Paroxysmal | 1 | (3.4%) | 4 | (8.2%) | 5 | (6.4%) |
Recurrent | 0 | (0%) | 1 | (2%) | 1 | (1.3%) |
Chronic | 0 | (0%) | 0 | (0%) | 0 | (0%) |
Atrial Fibrillation | 11 | (37.9%) | 9 | (18.4%) | 20 | (25.6%0 |
Paroxysmal | 8 | (27.6%) | 3 | (6.1%) | 11 | (14.1%) |
Recurrent | 0 | (0%) | 4 | (8.2%) | 4 | (5.1%) |
Chronic | 3 | (10.3%) | 1 | (2%) | 4 | (5.1%) |
|
The next table shows patient family history that was included in the specimen profiles and the relative breakdown between the control and test groups.
|
| Patients in | Patients in | Total |
| ICD Arm | Control Arm | Patients |
Patient Characteristics | (N = 29) | (N = 49) | (N = 78) |
|
|
Patient Family History (N, %) | | | | | | |
None | 20 | (69%) | 31 | (63.3%) | 51 | (65.4%) |
Long Q/T Syndrome | 0 | (0%) | 0 | (0%) | 0 | (0%) |
Grandparent | 0 | (0%) | 0 | (0%) | 0 | (0%) |
Parent | 0 | (0%) | 0 | (0%) | 0 | (0%) |
Sibling | 0 | (0%) | 0 | (0%) | 0 | (0%) |
Cousin | 0 | (0%) | 0 | (0%) | 0 | (0%) |
Sudden Cardiac Death | 5 | (17.2%) | 11 | (22.4%) | 16 | (20.5%) |
Grandparent | 0 | (0%) | 1 | (2%) | 1 | (1.3%) |
Parent | 4 | (13.8%) | 8 | (16.3%) | 12 | (15.4%) |
Sibling | 1 | (3.4%) | 3 | (6.1%) | 4 | (5.1%) |
Cousin | 0 | (0%) | 1 | (2%) | 1 | (1.3%) |
Sudden Death | 3 | (10.3%) | 4 | (8.2%) | 7 | (9%) |
Grandparent | 0 | (0%) | 0 | (0%) | 0 | (0%) |
Parent | 3 | (10.3%) | 3 | (6.1%) | 6 | (7.7%) |
Sibling | 1 | (3.4%) | 1 | (2%) | 2 | (2.6%) |
Cousin | 0 | (0%) | 0 | (0%) | 0 | (0%) |
Syncope | 2 | (6.9%) | 1 | (2%) | 3 | (3.8%) |
Grandparent | 0 | (0%) | 0 | (0%) | 0 | (0%) |
Parent | 2 | (6.9%) | 0 | (0%) | 2 | (2.6%) |
Sibling | 0 | (0%) | 1 | (2%) | 1 | (1.3%) |
Cousin | 0 | (0%) | 0 | (0%) | 0 | (0%) |
Deafness | 1 | (3.4%) | 4 | (8.2%) | 5 | (6.4%) |
Grandparent | 0 | (0%) | 1 | (2%) | 1 | (1.3%) |
Parent | 1 | (3.4%) | 2 | (4.1%) | 3 | (3.8%) |
Sibling | 0 | (0%) | 1 | (2%) | 1 | (1.3%) |
Cousin | 0 | (0%) | 0 | (0%) | 0 | (0%) |
History of Thrombo-embolic |
Event |
No | 24 | (82.8%) | 44 | (89.8%) | 68 | (87.2%) |
Yes | 59 | 17.2%) | 5 | (10.2%) | 10 | (12.8%) |
Time since most recent event | | | |
(years) |
Mean | 1.8 | 4.2 | 3.2 |
Standard Deviation | 1.4 | 6.3 | 4.7 |
Maximum-Minimum | 0.8-3.4 | 0.2-13.5 | 0.2-13.5 |
Type | | | | | | |
TIA | 2 | (6.9%) | 1 | (2%) | 3 | (3.8%) |
CVA | 2 | (6.9%) | 1 | (2%) | 3 | (3.8%) |
PE | 0 | (0%) | 1 | (2%) | 1 | (1.3%) |
Renal | 0 | (0%) | 0 | (0%) | 0 | (0%) |
Peripheral | 0 | (0%) | 0 | (0%) | 0 | (0%) |
Other | 1 | (3.4%) | 2 | (4.1%) | 3 | (3.8%) |
Other History |
History of Hyperthyroidism | 0 | (0%) | 2 | (4.1%) | 2 | (2.6%0 |
Hearing loss | 12 | (41.4%) | 16 | (52.7%) | 28 | (35.9%) |
|
The table below shows patient lifestyle characteristics that were included in the specimen profiles and the relative breakdown between the control and test groups.
|
| Patients in ICD | Patients in | Total |
| Arm | Control Arm | Patients |
Patient Characteristics | (N = 29) | (N = 49) | (N = 78) |
|
Does the Patient Smoke? | | | |
No | 23 (79.3%) | 42 (85.7%) | 65 (83.3%) |
Yes | 6 (20.7%) | 7 (14.3%) | 13 (16.7%) |
Number of years |
Mean | 41.6 | 39.6 | 40.4 |
Standard Deviation | 11.1 | 8 | 9 |
Maximum-Minimum | 30-55 | 30-50 | 30-55 |
Degree of Smoking |
1-2 packs a week | 1 (3.4%) | 2 (4.1%) | 3 (3.8%) |
3-5 packs a week | 0 (0%) | 1 (2%) | 1 (1.3%) |
5-10 packs a week | 2 (6.9%) | 3 (6.1%) | 5 (6.4%) |
10 or more packs a week | 1 (3.4%) | 0 (0%) | 1 (1.3%) |
Use of Alcohol |
No | 17 (58.6%) | 24 (49%) | 41 (52.6%) |
Yes | 12 (41.4%) | 25 (51%) | 37 (47.4%) |
Number of years |
Mean | 36.7 | 38.2 | 37.7 |
Standard Deviation | 17 | 13.9 | 14.7 |
Maximum-Minimum | 10-59 | 4-60 | 4-60 |
Degree of Drinking |
1-2 drinks a week | 5 (17.2%) | 6 (12.2%) | 11 (14.1%) |
3-5 drinks a week | 1 (3.4%) | 7 (14.3%) | 8 (10.3%) |
5-10 drinks a week | 4 (13.8%) | 6 (12.2%) | 10 (12.8%) |
10 or more drinks a week | 2 (6.9%) | 5 (10.2%) | 7 (9%) |
|
The following table shows patient baseline medications that were included in the specimen profiles and the relative breakdown between the control and test groups.
|
| Patients in | Patients in | Total |
| ICD Arm | Control Arm | Patients |
Patient Medications | (N = 29) | (N = 49) | (N = 78) |
|
|
Any Medications in Prior | | | | | | |
6 Months (N, %) |
No | 0 | (0%) | 0 | (0%) | 0 | (0%) |
Yes | 29 | (100%) | 49 | (100%) | 78 | (100%) |
Class I | 4 | (13.8%) | 1 | (2%) | 5 | (6.4%) |
Disopyramide | 0 | (0%) | 0 | (0%) | 0 | (0%) |
Flecainide | 0 | (0%) | 0 | (0%) | 0 | (0%) |
Mexiletine | 1 | (3.4%) | 0 | (0%) | 1 | (1.3%) |
Moricizine | 0 | (0%) | 0 | (0%) | 0 | (0%) |
Procainamide | 2 | (6.9%) | 0 | (0%) | 2 | (2.6%) |
Propafenone | 0 | (0%) | 1 | (2%) | 1 | (1.3%) |
Quinidine | 1 | (3.4%) | 0 | (0%) | 1 | (1.3%) |
Tocainide | 0 | (0%) | 0 | (0%) | 0 | (0%) |
Other | 0 | (0%) | 0 | (0%) | 0 | (0%) |
Class II | 14 | (48.3%) | 4 | (8.2%) | 18 | (23.1%) |
Amiodarone | 8 | (27.6%) | 2 | (4.1%) | 10 | (12.8%) |
Dofetilide | 0 | (0%) | 0 | (0%) | 0 | (0%) |
Sotalol | 7 | (24.1%) | 2 | (4.1%) | 9 | (11.5%) |
Other | 0 | (0%) | 0 | (0%) | 0 | (0%) |
Beta Blockers | 17 | (58.6%) | 36 | (73.5%) | 53 | (67.9%) |
Atenolol | 1 | (3.4%) | 9 | (18.4%) | 10 | (12.8%) |
Betaxolol | 0 | (0%) | 0 | (0%) | 0 | (0%) |
Bisoprolol | 0 | (0%) | 0 | (0%) | 0 | (0%) |
Bucindolol | 0 | (0%) | 0 | (0%) | 0 | (0%) |
Carvedilol | 4 | (13.8%) | 3 | (6.1%) | 7 | (9%) |
Metoprolol | 11 | (37.9%) | 22 | (44.9%) | 33 | (42.3%) |
Nadolol | 0 | (0%) | 0 | (0%) | 0 | (0%) |
penbutolol | 0 | (0%) | 0 | (0%) | 0 | (0%) |
Propanolol | 1 | (3.4%) | 2 | (4.1%) | 3 | (3.8%) |
Timolol | 0 | (0%) | 0 | (0%) | 0 | (0%) |
Other | 0 | (0%) | 1 | (2%) | 1 | (1.3%) |
Calcium Channel Blockers | 4 | (13.8%) | 10 | (20.4%) | 14 | (17.9%) |
Amlodipine | 2 | (6.9%) | 4 | (8.2%) | 6 | 97.7%) |
Diltiazem | 0 | (0%) | 3 | (6.1%) | 3 | (3.8%) |
Ibepridil | 0 | (0%) | 0 | (0%) | 0 | (0%) |
Felodipine | 0 | (0%) | 0 | (0%) | 0 | (0%) |
Nifedipine | 0 | (0%) | 3 | (6.1%) | 3 | (3.8%) |
Nisoldipine | 0 | (0%) | 0 | (0%) | 0 | (0%) |
Nimodipine | 0 | (0%) | 0 | (0%) | 0 | (0%) |
Verapamil | 1 | (3.4%) | 0 | (0%) | 1 | (1.3%) |
Other | 1 | (3.4%) | 0 | (0%) | 1 | (1.3%) |
Digoxin | 9 | (31%) | 3 | (6.1%) | 12 | (15.4%) |
Anti-Coagulants | 28 | (96.6%) | 46 | (93.9%) | 74 | (94.9%) |
Warfarin | 5 | (17.2%) | 8 | (16.3%) | 13 | (16.7%) |
Aspirin | 25 | (86.2%) | 40 | (81.6%) | 65 | (83.3%) |
Other | 4 | (13.8%) | 14 | (28.6%) | 18 | (23.1%) |
Other | 26 | (89.7%) | 39 | (79.6%) | 65 | (83.3%) |
|
Protein analysis from patient blood samples was carried out as described in the previous example. The CART (Classification and Regression Tree) method was used to identify class identifiers. The iterative partitioning algorithm used the test versus control groupings as the response variables and 2076 predictor variables that included 86 demographic variables and 1990 protein/peptide variables. Eligible protein/peptide variables were identified as peaks in spectral analysis of at least 4% of patients. Each patient's protein level for a given variable was averaged from two peak measurements.
FIG. 8 is the resulting tree analysis from the CART analysis. Tree analysis108 uses the five identified class identifiers P5, P6, P7, P8 and P9 to classify all patients, represented asgroup110, based on risk for fatal VT/VF.
The most effective class identifier is protein P5. Fifteen patients having P5 levels less than 0.0300685765 (measured in arbitrary units) were placed ingroup112. Thirteen, or 86.7%, were test patients with IMDs. Sixty-three patients having P5 levels greater than or equal to 0.0300685765 were placed ingroup114. Sixteen, or 25.4%, were test patients.
The class identifier shown tofurther partition group112 and represented as P6 was consumption of alcohol or lack of consumption for less than 20 years. Ten patients (eight of which had never consumed alcohol) were placed ingroup116. All ten patients, or 100%, were test patients. Five patients that had consumed alcohol for more than 20 years were placed ingroup118. Three of these five patients, or 60%, were test patients.
Group114 was then further partitioned based on levels of protein P7. Twenty patients having P7 levels greater than or equal to 0.1759674485 were placed intogroup120. All twenty, or 100%, were control patients. Forty-three patients having P7 levels less than 0.1759674485 were placed intogroup122. Twenty-seven of these forty-three patients, or 62.8%, were control patients.
Group122 was further partitioned based on levels of protein P8. Thirteen patients having P8 levels less than 0.314539267 were placed intogroup124. All thirteen, or 100%, were control patients. Thirty patients having P8 levels greater than or equal to 0.314539267 were placed into group126. Fourteen of these thirty patients, or 46.7%, were control patients.
Further partitioning of group126 was based on levels of protein P9. Thirteen patients having P9 levels of less than 0.0935425805 were placed intogroup128. Twelve of these patients, or 92.3%, were test patients. Seventeen patients having P9 levels greater than or equal to 0.0935425805 were placed intogroup130. Only four of these seventeen patients, or 23.5%, were test patients.
Thus, when applying tree analysis108, patients falling intogroups116 and128 have a significant risk of experiencing VT/VF and would benefit from an IMD. Conversely, patients falling intogroups120,124, and130 do not have a significant risk of experiencing VT/VF.
The table below summarizes the percentage of test patients belonging to each group of tree analysis108.
<0.0300685765 | Subtotal | | | | 15 | 86.7 |
| ≧20 years | | | | 5 | 60 |
| <20years | | | | 10 | 100 |
≧0.0300685765 | N/A | Subtotal | | | 63 | 25.4 |
| | ≧0.1759674485 | | | 20 | 0 |
| | <0.1759674485 | Subtotal | | 43 | 37.2 |
| | | <0.314539267 | | 13 | 0 |
| | | ≧0.314539267 | Subtotal | 30 | 53.3 |
| | | | ≧0.0935425805 | 17 | 23.5 |
| | | | <0.0935425805 | 13 | 92.3 |
Total | 78 | 37.2 |
|
*Each percentage is for the applicable group of the corresponding row; therefore the percentages do not sum to 100%, as they are calculated with different denominators (patient sample sizes). Columns one through five represent the class identifiers, P5-P9 and rows represent groups 112-130 obtained by using the class identifiers. Column 6 (Col. 6) is the total number of patients belonging to each corresponding group, and column 7 (Col. 7) is the percentage of patients in each group that are test patients. |
For example, to assess the percentage of test patients among all patients having P5 levels greater than or equal to 0.0300685765 and P7 levels less than 0.1759674485, begin at column 1 and select the row corresponding to ≧0.0300685765. Move to columns 2 and 3 (column 2 does not apply to these specific patients) and select the row in column 3 corresponding to <0.1759674485. Moving across to column 6, the number of patients having these class identifiers is 43, and the corresponding row in column 7 indicates that 37.2% of the 43 patients were test patients.
The following table summarizes the information regarding proteins P5, P7, P8, and P9.
|
| | | | Partitioning |
Molecular | | Fraction of | Spectrum | Peak |
Weight | Chip Type | Isolation | Range | Intensity |
|
|
P5 | Immobilized | Combined | High | 0.0300685765 |
11991 | Metal Affinity | fractions f2 | Protein |
| Surface | and f3 |
| | containing |
| | pH 5-7 |
P7 | Weak Cation | Fraction 1 | High | 0.1759674485 |
10552.4 | Exchange | containing | Protein |
| Surface | flow- |
| | through and |
| | pH = 9 |
P8 | Weak Cation | Fraction 1 | High | 0.314539267 |
43529.4 | Exchange | containing | Protein |
| Surface | flow- |
| | through and |
| | pH = 9 |
P9 | Hydrophobic | Fraction 1 | Protein | 0.0935425805 |
13806.8 | Surface | containing |
| | flow- |
| | through and |
| | pH = 9 |
|
The analysis resulted in four protein class identifiers and one demographic class identifier that correctly classifies patients based on risk of experiencing a true VT/VF episode.
The test procedures described above are not unique, nor are they necessarily the most efficient method of sorting patients who are candidates for an IMD from those that are not. Nevertheless, these procedures are illustrations of tests that can be used to screen patients to find out the ones who have a propensity for ventricular tachyarrhythmia and thus may be at increased risk of sudden cardiac death.
Depending upon the biochemical markers of interest, measurements of mass, concentration or abundance may be less important than determination of whether the marker is present or absent. The invention encompasses embodiments in which measurement of a biochemical marker in a patient includes determining whether the marker is present or not. For example, animal experimentation may establish that animals suffering sudden cardiac death exhibit an absence of a set of proteins and peptides having particular molecular weights. Similarly, animal experimentation may establish that animals suffering sudden cardiac death exhibit proteins or peptides that are otherwise not present. Detection of the presence or absence of such proteins or peptides in a human sample may have clinical significance, as the presence or absence proteins or peptides may be indicative of risk of sudden cardiac death.
In some cases, what is of interest is not the presence or absence of a biochemical marker, or its concentration on a single occasion, but an increase or decrease in the concentration or the rate of change, as demonstrated by two or more measurements separated by a time interval such as two weeks or one month. The invention supports consideration of a change as a basis for assessing a risk of ventricular tachyarrhythmia.
Test procedures such as the exemplary procedures described above can be automated, in whole or in part.FIG. 9 is an example of asystem132 that can perform an automated analysis of biochemical markers and can assess a risk of ventricular tachyarrhythmia in a patient as a function of the analysis.System132 includes asample input model134, which receives a sample for analysis, and ameasuring system136. In one embodiment of the invention,input module134 may include one or more biochips like those depicted inFIG. 5, and measuringsystem136 may comprise a SELDI-based mass analyzer. The invention is not limited to such components, however.
Aprocessor138 receives the measurements from measuringsystem136 and assesses a risk of ventricular tachyarrhythmia in the patient as a function by analyzing the measurements.Processor138 may apply a tree analysis, such as the analyses depicted inFIGS. 1,2,7, and8 to determine whether a patient is at risk of ventricular tachyarrhythmia.Processor138 may further assess a benefit of implanting a medical device in the patient as a function of the measurements, or administering an antiarrhythmic drug to the patient.
Anoutput module140 reports the results of the analysis.Output module140 may comprise a display screen, printer, or any other device that reports the results of the analysis. A benefit of implanting a medical device in the patient as a function of the measurement is assessed.
The invention may offer one or more advantages. Clinical data suggest that, in a significant number of cases, sudden cardiac death is the result of VT or VF. Episodes of VT or VF are treatable with an IMD or medication. The invention presents techniques for identifying the patients who are at risk of experiencing ventricular tachyarrhythmia. As a result, there is an improved chance that these patients will receive life-saving therapy, thereby reducing their risk of sudden cardiac death.
For example, recent evidence shows that VT/VF is treatable by administration of clonidine or vagal nerve stimulation, as well as through stimulation by an implantable cardioverter defibrillator (ICD). Thus, biomarkers may be used to identify patients that would benefit from these treatments and/or benefit from IMDs such as a drug pump to deliver intrathecal clonidine, a vagal nerve stimulator, or an ICD.
Therapies involving an IMD or medication need not be exclusive of one another. Furthermore, the invention supports therapies in addition to implantation of an IMD or regulation of a regimen of mediation. In some circumstances, the biomarkers may be more than symptomatic or indicative of the risk of VT or VF, and may be substantially causally related to the risk of VT or VF. In such circumstances, therapy may be directed to the biomarkers.
It may be possible, for example, to treat the patient by adjusting the concentration of biomarkers. When a concentration of certain protein biomarkers is found to be lower in a patient with VT or VF, then perhaps the patient can be treated by injecting those proteins into the blood, thereby restoring a more healthful concentration of the biomarkers. Conversely, when a concentration of certain protein biomarkers is found to be higher, then perhaps the patient can be treated by reducing the concentration of the protein biomarkers. A high concentration can be reduced by, for example, injection of enzymes that cleave or inhibit the activity of one or more protein biomarkers. Similarly, gene therapy can be used to alter protein and gene expression levels. Consequently, application of therapy may include determining one or more proteins or one or more genes, or a combination thereof, to be delivered to the patient.
The techniques of the invention may call for a sample from the patient. In many embodiments, the sample is one that is taken as a matter of course in a medical examination, such as a blood sample.
Further, the invention should reduce the incidents of false positives and false negatives. As a result, there is a better chance that patients that can benefit from an IMD will have a chance to receive an IMD. In addition, the invention includes the capability of being self-improving. As more clinical data are collected, different or more detailed tree analyses or other sorting techniques may be developed. Empirical experience may make tests more sensitive and more specific.
Various embodiments of the invention have been described. Various modifications can be made to the described embodiments without departing from the scope of the invention. For example, the invention is not limited to consideration of biochemical markers exclusively. The assessment of risk of ventricular tachyarrhythmia in the patient may also be a function of other measurable physiological factors. Electrophysiological measurements, such as an electrocardiogram, and hemodynamic factors, such as a measurement of ejection fraction, may be taken into consideration. Demographic factors such as the number of CABG procedures as well as alcohol and tobacco use are also factors that may be included.System110 inFIG. 9 may further include a sensor to measure a physiological factor, andprocessor116 may assess a risk of ventricular tachyarrhythmia as a function of the measurement of the physiological factor.
Although the invention has been described with proteins as biochemical markers, the invention is not limited to proteins. The invention also supports consideration of other markers, such as genetic markers, lipid markers and lipoprotein markers. The markers may be considered alone or in combination. For example, the invention supports risk assessments as a function of combinations of gene and protein markers. Techniques such as nuclear magnetic resonance, gene sequencing, or single nucleotide polymorphism (SNP) may be used to identify these markers. Consideration of markers such as these may result in enhanced sensitivity and specificity.
Analysis can e done using multiple techniques. In addition to generating a sorting tree, applying a logical analysis such as an IF-THEN statement, and artificial neural networks, one can assess a risk of ventricular tachyarrhythmia using linear clustering techniques (e.g., proximity, similarity, dissimilarity, weighted proximity, and principle component analysis) non-linear clustering techniques (e.g., artificial neural networks, Kohonen networks, pattern recognizers and empirical curve fitting), as well as logical procedures (e.g., CART, partition and hierarchical clustering algorithms). The invention is not limited to these techniques, however, and encompasses other linear analysis, non-linear analysis, logical analysis and conditional techniques.
Some of the techniques described above may be embodied as a computer-readable medium comprising instructions for a programmable processor such asprocessor138 inFIG. 9. The programmable processor may include one or more individual processors, which may act independently or in concert. A “computer-readable medium” includes but is not limited to read-only memory, Flash memory and a magnetic or optical storage medium.
Although the present invention has been described with reference to preferred embodiments, workers skilled in the art will recognize that changes may be made in form and detail without departing from the spirit and scope of the invention.