Improving Ribosomal RNA Integrity in Surgically Resected Human Brain Tumor Biopsies

Abstract

Background:

Biopsies extracted from brain cancer patients often display degraded ribosomal RNA, which makes them unusable in transcriptomic experiments. This has not been properly documented in previous works aimed at refining the molecular classification of brain cancer.

Objective:

To determine RNA integrity in a large cohort of human brain cancer biopsies and to evaluate different factors that may influence RNA integrity in both a murine model of glioblastoma and in additional subsets of patient biopsies.

Methods:

Total RNA was isolated from 255 biopsies of various human brain tumors (HBTs) and processed on a Bioanalyzer. Correct RNA integrity was considered for samples showing either the ribosomal 28S/18S peak ratio ≥1.2 or RNA integrity number ≥6. The time-dependent effect of ex vivo ischemia was evaluated in a murine model, whose results were tested in a new collection of 27 human biopsies. Multiple biopsy sampling was considered in a further set comprising 32 biopsies.

Results:

The 255 human biopsies revealed a substantial percentage of samples displaying degraded RNA (27.5%). The murine model confirmed the known relevance of ex vivo ischemia time in increased RNA degradation. Human biopsies extracted immediately after cauterization showed a trend toward less RNA degradation. Combining snap freezing and multiple sampling of biopsies, the percentage of patients with degraded RNA was reduced by twofold (15.6%).

Conclusions:

We provide a first concise study of factors influencing RNA degradation in HBT biopsies. Immediate biopsy removal after cauterization of the tumor area, snap freezing, and multiple sampling improve RNA quality.

Introduction

Studies based on gene expression microarrays revolutionized the oncology field by providing in single experiments a broader insight on transcript levels of explored cells or tissues than previous approaches based on investigation of one or few genes.^1–4 The field of neuro-oncology also experienced an impressive outbreak of works that refined the molecular features of histopathological types and proposed new subtypes.^5–8 Nowadays, the state-of-the-art technique for transcriptomic studies is massive RNA sequencing (RNA-seq), which has also contributed to an improved delineation of molecular traits in human brain tumors (HBTs).^9–11

However, most of those studies did not describe in adequate detail the integrity of RNA obtained from the samples. Usually, it is well stated in published articles that only biopsies with proper RNA integrity are used, but no information about the percentage of biopsies displaying degraded RNA is reported. While this issue has been commonly omitted, it is a crucial point to be addressed for any molecular tool to be implemented in clinical settings.

Commonly, parameters used to estimate the RNA quality are obtained using microcapillary-based electrophoretic systems, such as the Bioanalyzer and the Experion.^12,13 In short, there are two main parameters measured in this assay, the ratio of ribosomal RNAs 28S and 18S peak area (28S/18S ratio) and the RNA integrity number (RIN). RIN is a parameter proposed by Agilent that considers the entire electrophoretic profile for its estimation.¹⁴ On the contrary, the 28S/18S ratio is limited to the area under the peaks. It is stipulated that RNA samples display proper integrity when the 28S/18S ratio value is within the range of 1.2–1.5 or the RIN value is ≥5.^13–18

Previous studies have reported the proportion of fresh-frozen biopsies displaying degraded RNA. For instance, Micke et al. showed a high percentage (96%) of samples with good integrity RNA in a set of 47 samples comprising several cancer types and normal tissue.¹⁹ In another study comprising 37 fresh-frozen biopsies from prostate cancer, almost all samples (97%) displayed a RIN value above seven.²⁰ In contrast, Strand et al. observed a rate of biopsies with degraded RNA that ranged between 25% and 37.5%, depending on the criterion used, in a set of 24 invasive breast cancer samples.¹⁸ However, to our knowledge, there is no study addressing this topic in HBTs.

Some research teams have studied the factors leading to RNA degradation in biopsies. The elapsed time between surgical resection and freezing in liquid nitrogen is one of the main causes.^19,21 The physiological explanation for this effect is the induced ex vivo ischemia period during surgical resection. Hatzis et al. determined in 17 invasive breast cancers that 40 minutes of ex vivo ischemia time increased the RNA degradation in terms of ribosomal species, but it also affected the transcript level of some genes.²² In a study performed on 100 renal cell carcinoma samples, decreased RNA integrity was seen after 1 hour of ex vivo ischemia time.²¹ Nevertheless, Bao et al. did not observe an effect of ischemia time in a set of 51 colorectal fresh-frozen biopsies collected at different time periods.²³

Again, we have been unable to find previous studies focusing on factors affecting RNA degradation in biopsies of human brain cancers. For that reason, we aimed at evaluating the RNA degradation in a large set of 255 biopsies, and posteriorly, we assessed the effect of ex vivo ischemia time in a murine model and two subsets of human biopsies. Our results show that best RNA integrity results for brain tumor biopsies can be achieved by snap freezing after tumor cauterization and that multiple sampling is recommended due to tumor heterogeneity.

Methods

Collection of samples

Specific grant-associated biopsies

A total of 255 brain tumor biopsy samples were accrued during the eTUMOUR, HealthAgents, and MEDIVO2 projects between March 2004 and March 2009, accounting for several types and subtypes. These samples were collected for the most part not only at Hospital Universitari de Bellvitge IDI-BELL (n = 184) but also at hospitals associated with Centre Diagnòstic Pedralbes-Institut d'Altes Tecnologies (n = 59), Hospital Universitari Germans Trias i Pujol IDI-Badalona (n = 7), and Hospital Sant Joan de Déu (n = 5). All samples were collected in liquid nitrogen after surgical resection, except 33 biopsies that were also collected in RNAlater (Ambion; Thermo Fisher Scientific Inc.). In the latter case, the resected biopsy aliquot was immediately transferred to a 1.5-mL tube filled with RNAlater.

Biopsies before and after tumor resection

A set comprising 54 biopsies was collected before and after tumor resection. Specifically, the biopsy was resected from 27 patients immediately after delimiting the tumor area by cauterization to avoid an increased ischemia time. Then, a second specimen from each patient was drawn from the resected tumor mass to test the potential role of the surgery-related ischemia time in RNA degradation. This time was normally between 15 and 30 minutes. These biopsies were collected during the time period between January and May 2012 at the Hospital Universitari de Bellvitge IDI-BELL. Tumor types represented were glioblastoma (n = 8), metastasis (n = 6), meningothelial meningioma (n = 3), unreported diagnosis (n = 3), one anaplastic astrocytoma, one atypical meningioma, one fibrous meningioma, one transitional meningioma, one neuroblastoma of the adrenal gland and the sympathetic nervous system, one oligoastrocytoma, and one pleomorphic xanthoastrocytoma.

Triplicate biopsies per patient

The last set comprised 32 patients. Biopsies were collected between June 2012 and February 2015 at the Hospital Universitari de Bellvitge IDI-BELL. Surgeons extracted three biopsies from each patient immediately after cauterization (see Supplementary methods for the description of the procedure). Biopsies accounted for tumor types as follows: glioblastoma (n = 13), anaplastic oligoastrocytoma (n = 7), anaplastic astrocytoma (n = 2), grade II astrocytoma (n = 2), metastasis of melanoma (n = 2), and one patient for gliosarcoma, metastasis of lung adenocarcinoma, breast adenocarcinoma, breast carcinoma, anaplastic oligodendroglioma, and grade II oligoastrocytoma. The protocol agreed to with the surgeons is available as Supplementary Data.

Storage and histopathological analysis of samples

All biopsies were placed in cryotubes and immediately snap-frozen in liquid nitrogen after surgical removal, unless otherwise indicated. Tumor samples not used for RNA extraction were fixed in 4% buffered formalin and embedded in paraffin. For routine histological examination, 4-μm-thick sections were stained with hematoxylin and eosin (HE). Given that samples were collected from 2004 to 2015, the WHO 2000 and 2007 Nervous System Classification criteria were used for diagnosis.^24,25

The local Ethics Committee approved the collection of the biopsies described above and informed consent was obtained from all patients.

RNA isolation

Total RNA was isolated using the mirVana miRNA Isolation Kit (Ambion; Thermo Fisher Scientific Inc.), following the manufacturer's instructions. Approximately 50 mg of frozen material was split from the entire biopsy using a mortar, which was previously chilled in liquid nitrogen. For small biopsies, all of the available material was used. RNA yield per biopsy is described in Supplementary Tables (Supplementary Data are available online at www.liebertpub.com/bio).

Isolated RNA was characterized using a NanoDrop spectrophotometer (NanoDrop Technologies). The absence of protein contamination was monitored by the 260/280 nm ratio of absorbance. RNA samples with a ratio ranging between 1.6 and 2.3 were accepted for further processing, as specified in the eTUMOUR project.²⁶

RNA integrity was assessed using the capillary electrophoretic system 2100 Bioanalyzer (Agilent). Only RNA samples producing a 28S/18S ratio equal or higher than 1.2 or a RIN number equal or higher than 6 were selected, as specified in the eTUMOUR project.²⁶ Measured values of 28S/18S ratio and RIN are available in Supplementary Tables S1–S3. Currently, additional approaches to assess RNA integrity apart from the abovementioned ones based on ribosomal species can be tested, such as the ones based on quantitative polymerase chain reaction (qPCR). However, 28S/18S ratio and RIN were the only measurements agreed to in those projects providing the most biopsies for this study. Thus, the use of extensive qPCR assays for RNA integrity is unfortunately beyond the scope of this work.

qPCR of the gene LGALS1 on a subset of samples

Although the entire study was designed to assess the integrity of RNA through the measurement of ribosomal RNAs, an additional evaluation was performed by qPCR of the LGALS1 gene. Duplicate biopsies from the set of patients with pre- and postcauterization specimens were assessed. They represented two pairs of samples with correct RNA integrity of the precauterization biopsy, one pair with correct integrity of the postsurgery biopsy and one pair displaying degraded RNA in both biopsies. qPCR experiments were performed using triplicate measurements per RNA sample as previously described.²⁷

Generation of a murine model to simulate ex vivo normal body temperature ischemia in brain tumor samples

Animals and cells

A total of 29 C57BL/6 female mice, with a weight range of 20–23 g, were used in this study. Animals were obtained from Charles River Laboratories (France) and housed at the animal facility of the Universitat Autònoma de Barcelona (UAB). All animal studies were approved by the local ethics committee according to regional and state legislations (protocol DARP-3255/CEEAH-530). GL261 mouse glioma cells were obtained and cultured exactly as described by Simões et al.²⁸

Inoculation of the mouse brain with GL261 tumor glial cells

Tumors were induced in 29 mice by intracranial stereotactic injection of 10⁵ GL261 cells in the caudate nucleus, as previously described.^28,29 About 15 minutes after being given a dose of analgesia (Meloxicam subcutaneous, s.c., 1.0 mg/kg), animals were anesthetized (Ketamine-Xylazine, 80–10 mg/kg intraperitoneal, i.p.) and then immobilized in a stereotactic holder (Kopf Instruments). The skull was exposed and a high-speed microdriller (Fine Science Tools) was used to make a small hole in its surface (1 mm): 2.3 mm to the right of the midline, as measured from the Bregma. A 26G Hamilton syringe (Hamilton), positioned on a digital push-pull microinjector (KD Scientific), was advanced through this hole, 2.3 mm from the cortical surface into the striatum, to deliver 10⁵ GL261 cells (in 4 μL RPMI medium) at a rate of 2 μL/min. The syringe was slowly removed 3–5 minutes after the injection was finished and the scission site closed with suture silk (5.0). Animals were left to recover from anesthesia in a warm environment (∼25°C) and, as they began to wake up, a stronger analgesic (opioid) was given: Buprenorphine s.c., 0.1 mg/kg. Meloxicam analgesia was repeatedly administrated at 24 and 48 hours/postsurgery.

Animal sacrifice and encephalon removal

Animals were sacrificed by an intraperitoneal injection of sodium pentobarbital (60 mg/mL) at a dose of 200 mg/kg and posterior cervical dislocation. Once the animal became unresponsive to mechanical foot stimulation, posterior cervical dislocation was performed and the head sectioned from the rest of the body with sterile scissors. The upper part of the skull was then removed by an incision at each occipital condyle, which was followed by a dissection in the anterior direction up to the nasal cavity. The encephalon was removed by lifting it up with sterile dissection tweezers.

Dissection of the tumor mass

The hemisphere inoculated with tumor cells, together with a thin surrounding layer of the parental hemisphere, was separated from the rest of the brain by using a sterile scalpel and dissection tweezers. The cerebral parenchyma was progressively removed to keep only the tumor mass, characterized by its more mucous appearance with respect to normal brain. In the case of necrotic specimens, the tumor mass was clearly identifiable due to its dark color. In contrast, the identification of non-necrotic tumor masses was more difficult since their color was very similar to the nontumor parenchyma. However, it displayed a more mucous appearance compared with the normal brain parenchyma.

The whole procedure described in this section was performed at room temperature. The time elapsed from animal death until tumor removal ranged between 5 and 7 minutes.

Simulation of ex vivo normal body temperature ischemia

Simulation of ex vivo warm ischemia was performed in 4 mice displaying evident necrotic tumor masses and in 14 mice showing little necrosis or non-necrotic tumors. Immediately after tumor mass resection, an aliquot (<1 mm³) was submerged into formalin for posterior histological verification. Such validation was performed using standard protocols (paraffin-embedded and HE-stained tissue slides).

To simulate body temperature, tumor masses were introduced into separate 1.8-mL cryotubes prefilled with phosphate-buffered saline (PBS) at 37°C. Samples were incubated at 37°C for 30 minutes and snap-frozen in liquid nitrogen after this period. Furthermore, 7 of the 14 non-necrotic mice tumors were incubated for 15 minutes instead of 30 minutes. Remaining tumor specimens were not subjected to body temperature incubation and were snap-frozen in liquid nitrogen immediately after dissection.

Statistical analysis

The difference of means was assessed through a t-test using the t.test function available in R for the large dataset comprising 255 biopsies. For remaining samples, this assessment was performed using the Wilcoxon test using the R software with available wilcox.test function³⁰ due to the lack of normal distribution of their values.

The difference of proportions between two groups was assessed using the prop.test function in R, which performs Pearson's chi-squared test. For the set of triple biopsies, a logistic regression model using the binomial distribution as a link function was fitted to evaluate the dependency of tumor types and surgeons on the RNA integrity (degraded or nondegraded RNA).

A linear model was fitted to either 28S/18S ratios or RIN values to determine the involvement of tumor type in RNA integrity. For this purpose, the lm available in the stats R package was used. In the same package, the function, cor, was used to compute the correlation between micrograms of isolated RNA and the two parameters mentioned above.

Results

High percentage of collected brain biopsies displayed elevated RNA degradation

The first step of the study consisted of assessing whether the collection medium for biopsies played a role in the degradation of RNA. For that purpose, we collected a set of 33 biopsies in two aliquots, one in liquid nitrogen (the method usually performed by our team) and the second one in RNAlater. As it can be seen in Figure 1A, B, no difference in RNA integrity was detected between collecting biopsies in RNAlater compared with snap-frozen in liquid nitrogen. These results indicated that the collection medium was not responsible for RNA degradation, and we chose liquid nitrogen as the collection medium in the remainder of this study.

FIG. 1.

Characterization of RNA integrity in human brain tumor samples. (A) Comparison of 28S/18S ratio for 33 biopsies simultaneously collected in liquid nitrogen and RNAlater at the surgery room is depicted in this boxplot. As indicated in the text, no significant difference was detected. (B) The same as (A), but using the RIN as measure of RNA integrity. (C) This figure shows the average 28S/18S ratio values for those biopsies hybridized and nonhybridized in Affymetrix microchips. Statistical difference between means was also computed using a two-sided t-test. (D) The percentage of samples per tumor type was plotted for nonhybridized biopsies. (E) The same as (D), but using those hybridized biopsies. A, low-grade astrocytoma; AA, anaplastic astrocytoma; AEP, anaplastic ependymoma; AMG, atypical meningioma; ANG, angioma; AOA, anaplastic oligoastrocytoma; AOD, anaplastic oligodendroglioma; GB, glioblastoma; HMB, hemangioblastoma; HYA, hypophysis adenoma; LMFB, B lymphoma; M, metastasis; MG, meningioma; ND, no diagnosis; NEURC, neurocytoma; OA, low-grade oligoastrocytoma; OD, low-grade oligodendroglioma; PA, pilocytic astrocytoma; SCW, schwanoma; T, teratoma; TMG, transitional meningioma; RIN, RNA integrity number.

The total number of biopsies collected within the eTUMOUR project by the UAB partner was 255. Of these, 27.5% (70) showed a 28S/18S ribosomal peak ratio or RIN that did not meet the standards of RNA integrity agreed to in the eTUMOUR project (Fig. 1C). This percentage of biopsy samples, which were not suitable for microarray hybridization, was detected since the beginning of the project and it was independent of the tumor type (Fig. 1D, E). This is also supported by the absence of statistical difference based on the chi-squared test observed in the pairwise comparison of matching tumor types between RNA samples from biopsies accepted and nonaccepted (RNA quality noncompliant) for hybridization. In this respect, the average values of the 28S/18S ratio and RIN per tumor type also support the lack of correlation among specific tumor types and RNA degradation status (Table 1). Only low-grade oligodendroglioma, meningioma, and metastasis displayed a significantly lower value of 28S/18S ratio or RIN. Furthermore, the absence of correlation between these two parameters with micrograms of RNA isolated was observed (Cor_28S/18S-μg = 0.10 and Cor_RIN-μg = 0.14).

Table 1.

RNA Degradation by Tumor Type

	28S/18S			RIN			Isolated RNA (μg)
	Mean	SD	p	Mean	SD	p	Mean	SD
Anaplastic astrocytoma (27)	1.23	0.61	Ref.	7.2	2.0	Ref.	33.2	25.5
Anaplastic oligoastrocytoma (29)	1.32	0.42	0.56	7.2	1.5	0.90	30.0	32.4
Anaplastic oligodendroglioma (13)	1.15	0.62	0.71	6.9	2.4	0.80	59.0	45.8
Atypical meningioma (17)	1.14	0.52	0.62	6.6	2.5	0.53	97.5	74.4
Glioblastoma (128)	1.02	0.60	0.09	6.2	2.5	0.08	46.9	40.2
Grade II astrocytoma (6)	1.47	0.35	0.36	7.3	0.5	0.89	16.1	16.3
Grade II oligoastrocytoma (3)	1.30	0.26	0.84	7.5	1.1	0.83	25.2	16.7
Low-grade oligoastrocytoma (8)	1.33	0.58	0.67	6.9	2.0	0.82	24.5	12.6
Low-grade astrocytoma (16)	1.33	0.47	0.59	6.1	2.2	0.25	35.0	29.0
Low-grade oligodendroglioma (7)	0.70	0.70	0.03	5.6	2.6	0.28	30.3	26.2
Meningioma (47)	1.03	0.68	0.16	5.8	2.7	0.04	49.4	32.4
Metastasis (28)	1.03	0.59	0.16	5.7	2.6	0.02	80.7	57.7
Pilocytic astrocytoma (3)	1.07	0.31	0.65	7.6	1.8	0.78	26.2	34.1
Transitional meningioma (7)	1.29	0.37	0.81	6.8	2.6	0.75	70.4	49.6

The mean and standard deviation (SD) of 28S/18S ratio, RIN, and total isolated RNA are provided per those tumor types represented at least by three biopsies. The p-value provides the probability that the mean of each tumor type differs from the reference group (Ref., anaplastic astrocytoma). The number of biopsies per tumor type is indicated within brackets. As duplicate and triplicate biopsies per patient (Supplementary Tables S2 and S3) were used to build the table, multiple biopsies may correspond to one single patient.

RIN, RNA integrity number.

Therefore, we hypothesized that the time elapsed from tumor resection until snap freezing of biopsies could be critical. Thus, we generated a murine model of glioblastoma to evaluate the effect of ex vivo ischemia at normal body temperature in RNA from brain tumors.

Ex vivo normal body temperature ischemia in a murine model increases RNA degradation in non-necrotic samples

Incubation for 30 minutes in PBS at 37°C did not produce differences in RNA degradation for necrotic tumors compared with necrotic tumors snap-frozen in liquid nitrogen (Fig. 2A, B). In contrast, non-necrotic tumors showed a statistically significant decrease of the 28S/18S ratio for cases incubated for both 15 and 30 minutes in PBS at 37°C (Fig. 2C, D).

FIG. 2.

Ex vivo simulation of ischemia at normal body temperature in a murine model of glioblastoma. Effect of normal body temperature ischemia on RNA degradation for both necrotic and non-necrotic mice tumors is shown. (A, B) Box plots of both 28S/18S ratio and RIN number, respectively, for necrotic tumors are shown. (C, D) Average 28/18S ratio and RIN values for non-necrotic tumors at three time points (snap-frozen, 15 and 30 minutes) are shown. Difference between the 28S/18S ratio or RIN value of snap-frozen samples and those ones incubated for 15 or 30 minutes at 37°C was assessed using a Wilcoxon test.

From our results, RNA degradation as measured from the 28S/18S ratio and RIN occurs for non-necrotic tumor specimens when simulating an ex vivo normal body temperature ischemia period. This phenomenon is not observed in necrotic tumors. Translating these findings into the routine clinical practice, the 25%–30% of human biopsies showing RNA degradation, and unsuitable for microarray hybridization, could be explained by an increased ex vivo ischemia time.

Fast resection of biopsies improves RNA integrity

Standard surgical practice during HBT removal requires surgeons halting blood flow to the tumor by cauterization when delimiting the tumor mass. This results in variable ischemia time at body temperature before biopsy removal (between 15 and 30 minutes).

We evaluated, in a new set of human biopsies, the RNA integrity of 27 samples before and after tumor resection (the procedure undertaken for the 255 biopsies described above). The percentage of biopsies displaying acceptable RNA integrity was similar in those ones taken before and after tumor resection (59.3% [16/27] and 55.6% [15/27], respectively). However, looking at the 28S/18S ratio and RIN values in more detail, the average values were always higher, although not significantly, in samples immediately resected after cauterization than in the ones sampled at the end of the surgery, both in terms of 28S/18S (pre = 1.02 ± 0.48, post = 0.88 ± 0.49, p = 0.335) and RIN values (pre = 6.09 ± 2.21, post = 5.84 ± 2.70, p = 0.843).

Furthermore, we tested whether the detectable gene expression level of LGALS1 was related to the integrity of ribosomal RNA, according to the predefined quality thresholds. For that, four pairs of biopsies extracted before and after cauterization in four different patients were evaluated (Supplementary Table S4). The expression level of LGALS1 in those biopsies displaying correct ribosomal RNA integrity (double-02-pre, double-08-pre, and double-24-post) was at least three times higher than in the other pair (double-02-post, double-08-post, and double-24-pre). That is, a fold change^pre/post ≥3 was noticed for biopsies of patients, double-02 and -08, while a fold change^pre/post ≤1/3 was seen for patient double-24. The patient with degraded rRNA in both biopsies (double-19) showed twice the expression level in the postcauterization biopsy. Nevertheless, the Ct value obtained from those two biopsies was very similar to the other biopsies displaying degraded ribosomal RNA (double-02-post, double-08-post, and double-24-pre) with values close to or above 25, while Ct values from samples with correct RNA integrity (Ct^double-02 = 22.1, Ct^double-08 = 20.6, and Ct^double-24 = 22.2) were similar to the value observed in a previous work,²⁷ in which the average Ct value was 21.9 ± 1.2.

These results suggest the possible advantage of taking the biopsy as the first step after cauterization of the tumor mass so that the likely effects of ischemia during the operation are avoided and maximal RNA integrity is preserved. Moreover, these data also reveal the improvement in the approach of surgeons in limiting the number of biopsies with degraded RNA. That is, if we consider the 27 biopsies collected in this subset, only 22.2% (6 of 27) had degraded RNA, which improves the rate obtained in the set of 255 biopsies mentioned above (27.5%).

Multiple sampling per patient increases the percentage of biopsies with acceptable RNA integrity

Given that the integrity of the RNA of the tissue does not vary significantly with the moment it is obtained during the surgical procedure, we hypothesized that other factors may be contributing to the RNA integrity of biopsies. The intratumor heterogeneity of cancer tissue is well known in general and particularly in brain tumors such as gliomas.³¹ We tested whether, by sampling several biopsies per patient, the chance of having at least one biopsy with proper RNA integrity increased.

To evaluate that hypothesis, surgeons resected three biopsies per tumor mass in 32 patients. Biopsies were resected immediately after tumor cauterization to maintain possible beneficial effects in RNA quality, as demonstrated in the previous section. As shown in Table 2, half of the patients harbored nondegraded RNA and just five (15.6%) had all three biopsies displaying degraded RNA. Although this percentage is lower than the one observed in the two previous sets, there are still a non-negligible number of samples not fulfilling the standards of RNA integrity.

Table 2.

RNA Degradation in Triple Biopsy Sampling per Patient

No. of biopsies deg. RNA/patient	No. of patients	Patients (%)
0	16	50.0
1	5	15.6
2	6	18.8
3	5	15.6

This table shows the number (number of patients) and percentage (% of patients) with none, one, two, or three biopsies with degraded RNA (number of biopsies deg. RNA/patient).

Although we had already observed that the tumor type is not related with RNA integrity, we fitted a logistic regression model using the binomial distribution as a function link and the tumor type as predictor variable. No tumor type had a significantly higher chance of having three degraded biopsies per patient compared with grade II astrocytoma, which was set as the reference tumor according to alphabetical order. Actually, all p-values were close to one (data not shown). A similar result was obtained when assessing the involvement of surgeons who had extracted the biopsies.

Considering that biopsies were sequentially extracted, we computed the percentage of patients with at least one biopsy having nondegraded RNA observed in the order of extraction. Looking at the first biopsy removed from each patient, seven patients (21.9%) had biopsies displaying degraded RNA, which is a similar percentage to the one obtained in the previous section. By extracting two biopsies, the number of patients with degraded RNA dropped to five (15.6%), which is the same figure as the one obtained after the extraction of the third biopsy.

Discussion

RNA integrity is a main issue in oncology when isolating RNA from a biopsy for subsequent molecular analyses. Acceptable RNA integrity has been mandatory to conduct reliable and reproducible gene expression microarray experiments during the past decade.^14,16,17 Nowadays, this requirement still prevails for RNA-seq experiments.^32,33 Therefore, any study aiming to establish a biomarker based on transcriptomics relies on the integrity of RNA for the proposal to be trustworthy.

In a large dataset of brain tumor biopsies, we observed a high percentage, almost 30%, of biopsies displaying RNA degradation. This was a fatidic concern given that the aim of this type of project was to identify biomarkers for different types and subtypes of brain tumors. That is, any diagnostic or prognostic assay with such a high rate of failure would not be implementable in standard clinical settings.

For that reason, we investigated the best approach for biopsy collection in the surgery room so that we could maximize the number of specimens with proper RNA integrity and thus render them usable for diagnosis and prognosis. The first question that we addressed was the involvement of ex vivo normal body temperature in the stability of RNA. The rationale was to reproduce the likely ischemic conditions that happen during surgery between the cauterization of the tumor tissue and the removal of the biopsy. For this purpose, we evaluated in a murine model of glioblastoma whether incubating the extracted biopsy in PBS medium at 37°C for 15 or 30 minutes increased the RNA degradation compared with snap-frozen biopsies after resection. Indeed, there was a decrease in RNA integrity with increased incubation time. However, that effect was only seen in non-necrotic tumors (Fig. 2). Lack of such an effect in necrotic tumors may be explained by the activation of hypoxia-related metabolism, which is well established as a process that is led by the expression of hypoxia-inducible factor 1 (HIF1).^34–36 HIF1 is a transcription factor for several genes, whose expression counterbalances the effects of lack of oxygen supply. The HIF1-induced pathway in necrotic tumors could be already activated before surgical biopsy resection. Thus, cells of these tumors would be protected against acute damage, such as massive ribosomal RNA degradation, compared with non-necrotic specimens. This is the most plausible hypothesis that we can postulate, although its demonstration is beyond the goals of this study. Although the murine model is not totally comparable with humans, this experiment highlighted the relevance of extracting a biopsy immediately after cauterization of the tumor mass.

The promising results with the murine model led us to evaluate that effect in humans. In a set of 27 patients, we compared the RNA integrity of a first biopsy extracted immediately after cauterization and a second one taken after the complete resection of the tumor. As a result, the integrity of biopsies extracted without elapsed ischemic time was not statistically different from those resected later on. Nevertheless, 28S/18S and RIN average values were consistently higher in those specimens initially removed, and the percentage of biopsies with proper RNA integrity decreased compared with the initial set of 255 biopsies (22.2% vs. 27.5%). In addition, the expression level of LGALS1 was more than 10 times higher in two precauterization biopsies than in their postcauterization counterparts (fold change^pre/post-double-02 = 11.7 and fold change^pre/post-double-08 = 50.4). Regardless, one of the postcauterization biopsies showed higher integrity of ribosomal RNA than the precauterization one (fold change^pre/post-double-24 = 0.3), which may be related to the known heterogeneity of HBTs.³¹ The qPCR experiments performed confirmed that degraded ribosomal RNA implies degraded messenger RNA.

In light of these results, we speculated about additional factors potentially involved in RNA integrity. Provided that brain tumors, and specially gliomas and glioblastoma, are very heterogeneous, we hypothesized that apart from an immediate extraction of the biopsy, several specimens per patient should be taken to ensure at least one biopsy with nondegraded RNA. Interestingly, the percentage of patients displaying degraded RNA in all biopsies dropped to 15.6% (5 of 32) compared with the set of biopsies extracted after cauterization and resection (22.2%, 6 of 27). Nonetheless, the rate of unusable biopsies for molecular analyses remained prominent.

Although the type of tumor was not statistically associated with RNA integrity, all patients with three degraded biopsies suffered from high-grade tumors (three glioblastomas, one gliosarcoma, and one metastasis of breast adenocarcinoma). The lack of significance is likely to be related to the small sample size, and it cannot be ruled out that tumor type may also play a role in RNA integrity. However, high-grade brain tumors are the most commonly observed and there was also absence of significance in the set of 255 biopsies of this study. At any rate, data from our study confirm that this is an issue deserving further attention. For instance, the average 28S/18S and RIN values in one of the least aggressive tumors, meningiomas, were similar or even lower than the ones in glioblastomas, the most aggressive HBT (Table 1).

It is well known that gliomas shift their metabolism toward an increased glycolytic activity^37,38 and are often exposed to fluctuating hypoxic conditions.³⁹ This may imply a metabolic reorganization of tumor cells with consequences on ribosomal RNA turnover and integrity. In this respect, a fast increase in cytoplasmic turnover of ribosomes in yeast has been demonstrated upon mTOR inhibition conditions mimicking starvation.⁴⁰ A similar situation may be encountered under fluctuating hypoxia conditions in tumor cells.

Therefore, it appears plausible that in addition to factors identified in this study (speed of extracting the biopsy and multiple sampling), the metabolic stage of a tumor specimen, which is often undetermined before surgery, could be a relevant feature for RNA integrity.

To the best of our knowledge, this is the first work investigating in detail the causes of the high percentage of brain tumor biopsies displaying degraded RNA. Subsequent to obtaining a percentage close to 30% of degradation in the 255 human biopsy dataset, we have shown in a murine model of glioblastoma that an ex vivo ischemia time of 15 minutes increases the RNA degradation. Further collection and analysis of human biopsies have shown that (1) biopsies should be snap-frozen early after cauterization to restrict RNA degradation and that (2) multiple sampling is recommended to increase the chances of obtaining at least one biopsy with proper RNA integrity per patient.

The lowest rate of RNA degradation achieved in this work was 15.6%, which still remains too elevated whenever the final purpose is to use samples for routine diagnosis. Further studies should focus on linking the metabolic stage of tumors and the integrity of their RNA. In this respect, the fluorescence-guided multiple sampling scheme for tissue collection described by Sottoriva et al.³¹ could provide a relevant avenue for further investigation.

Footnotes

Acknowledgments

The authors thank Dr. Juan José Acebes, former head of the Neurosurgery Ward at Hospital Universitari de Bellvitge, for guidance and support during the initial phase of this study. Participation in biopsy and clinical data accrual by Dr. Angel Moreno and Dr. Jesús Pujol (Centre Diagnòstic Pedralbes-Institut d'Altes Tecnologies), Dr. Jaume Capellades (Hospital Universitari Germans Trias i Pujol IDI-Badalona), and Dr. Antoni Capdevila (Hospital Sant Joan de Déu) is also acknowledged. This work was funded by the EU-funded grant eTUMOUR (FP6-2002-LIFESCIHEALTH 503094) and the Spanish grants, MARESCAN (SAF2011-23870) and MOLIMAGLIO (SAF2014-52332-R), from MINECO. Additional funding was received from CIBER-BBN (Centro de Investigación Biomédica en Red–Bioingeniería, Biomateriales y Nanomedicina []), an initiative of the Instituto de Salud Carlos III (Spain) cofunded by EU FEDER funds.

Author Disclosure Statement

No conflicting financial interests exist.

References

Chari

, Thu

, Wilson

, et al. Integrating the multiple dimensions of genomic and epigenomic landscapes of cancer. Cancer Metastasis Rev, 2010; 29:73–93.

Nannini

, Pantaleo

, Maleddu

, Astolfi

, Formica

, Biasco

Gene expression profiling in colorectal cancer using microarray technologies: Results and perspectives. Cancer Treat Rev, 2009; 35:201–209.

Weigelt

, Baehner

, Reis-Filho

. The contribution of gene expression profiling to breast cancer classification, prognostication and prediction: A retrospective of the last decade. J Pathol, 2010; 220:263–280.

, He

. Molecular classification of non-small-cell lung cancer: Diagnosis, individualized treatment, and prognosis. Front Med, 2013; 7:157–171.

Gravendeel

LAM

, Kouwenhoven

MCM

, Gevaert

, et al. Intrinsic gene expression profiles of gliomas are a better predictor of survival than histology. Cancer Res, 2009; 69:9065–9072.

, Walling

, Ahn

, et al. Unsupervised analysis of transcriptomic profiles reveals six glioma subtypes. Cancer Res, 2009; 69:2091–2099.

Parsons

, Jones

, Zhang

, et al. An integrated genomic analysis of human glioblastoma multiforme. Science, 2008; 321:1807–1812.

Phillips

, Kharbanda

, Chen

, et al. Molecular subclasses of high-grade glioma predict prognosis, delineate a pattern of disease progression, and resemble stages in neurogenesis. Cancer Cell, 2006; 9:157–173.

Bao

Z-S

, Chen

H-M

, Yang

M-Y

, et al. RNA-seq of 272 gliomas revealed a novel, recurrent PTPRZ1-MET fusion transcript in secondary glioblastomas. Genome Res, 2014; 24:1765–1773.

10.

Patel

, Tirosh

, Trombetta

, et al. Single-cell RNA-seq highlights intratumoral heterogeneity in primary glioblastoma. Science, 2014; 344:1396–1401.

11.

Shah

, Lankerovich

, Lee

, Yoon

J-G

, Schroeder

, Foltz

. Exploration of the gene fusion landscape of glioblastoma using transcriptome sequencing and copy number data. BMC Genomics, 2013; 14:818.

12.

Copois

, Bibeau

, Bascoul-Mollevi

, et al. Impact of RNA degradation on gene expression profiles: Assessment of different methods to reliably determine RNA quality. J Biotechnol, 2007; 127:549–559.

13.

Imbeaud

, Graudens

, Boulanger

, et al. Towards standardization of RNA quality assessment using user-independent classifiers of microcapillary electrophoresis traces. Nucleic Acids Res, 2005; 33:e56.

14.

Shi

, Perkins

, Fang

, Tong

. Reproducible and reliable microarray results through quality control: Good laboratory proficiency and appropriate data analysis practices are essential. Curr Opin Biotechnol, 2008; 19:10–18.

15.

Atz

, Walsh

, Cartagena

, et al. Methodological considerations for gene expression profiling of human brain. J Neurosci Methods, 2007; 163:295–309.

16.

Mane

, Evans

, Cooper

, et al. Transcriptome sequencing of the Microarray Quality Control (MAQC) RNA reference samples using next generation sequencing. BMC Genomics, 2009; 10:264.

17.

Shi

, Campbell

, Jones

, et al. The MicroArray Quality Control (MAQC)-II study of common practices for the development and validation of microarray-based predictive models. Nat Biotechnol, 2010; 28:827–838.

18.

Strand

, Enell

, Hedenfalk

, Fernö

. RNA quality in frozen breast cancer samples and the influence on gene expression analysis—A comparison of three evaluation methods using microcapillary electrophoresis traces. BMC Mol Biol, 2007; 8:38.

19.

Micke

, Ohshima

, Tahmasebpoor

, et al. Biobanking of fresh frozen tissue: RNA is stable in nonfixed surgical specimens. Lab Invest, 2006; 86:202–211.

20.

Brimo

, Aprikian

, Latour

, et al. Strategies for biochemical and pathologic quality assurance in a large multi-institutional biorepository; The experience of the PROCURE Quebec Prostate Cancer Biobank. Biopreserv Biobank, 2013; 11:285–290.

21.

Sun

, Sun

, Hao

, et al. Effect of duration of ex vivo ischemia time and storage period on RNA quality in biobanked human renal cell carcinoma tissue. Ann Surg Oncol, 2016; 23:297–304.

22.

Hatzis

, Sun

, Yao

, et al. Effects of tissue handling on RNA integrity and microarray measurements from resected breast cancers. J Natl Cancer Inst, 2011; 103:1871–1883.

23.

Bao

W-G

, Zhang

J-G

, et al. Biobanking of fresh-frozen human colon tissues: Impact of tissue ex-vivo ischemia times and storage periods on RNA quality. Ann Surg Oncol, 2013; 20:1737–1744.

24.

Kleihues

, Cavenee

. World Health Organization Classification of Tumours: Pathology and Genetics: Tumours of the Nervous System. Lyon: IARC press; 2000: 1–8.

25.

Louis

, Ohgaki

, Wiestler

, et al. The 2007 WHO classification of tumours of the central nervous system. Acta Neuropathol, 2007; 114:97–109.

26.

Julià-Sapé

, Lurgi

, Mier

, et al. Strategies for annotation and curation of translational databases: The eTUMOUR project. Database (Oxford), 2012; 2012:bas035.

27.

Castells

, Acebes

, Majós

, et al. Robustness of equations that define molecular subtypes of glioblastoma tumors based on five transcripts measured by RT-PCR. OMICS, 2015; 19:41–51.

28.

Simões

, García-Martín

, Cerdán

, Arús

. Perturbation of mouse glioma MRS pattern by induced acute hyperglycemia. NMR Biomed, 2008; 21:251–264.

29.

Simões R

, Delgado-Goñi

, Lope-Piedrafita

, Arús

. 1H-MRSI pattern perturbation in a mouse glioma: The effects of acute hyperglycemia and moderate hypothermia. NMR Biomed, 2010; 23:23–33.

30.

R Development Core Team. R: A Language and Environment for Statistical Computing. Vienna, Austria: R Foundation for Statistical Computing; 2015.

31.

Sottoriva

, Spiteri

, Piccirillo

SGM

, et al. Intratumor heterogeneity in human glioblastoma reflects cancer evolutionary dynamics. Proc Natl Acad Sci U S A, 2013; 110:4009–4014.

32.

Gallego Romero

, Pai

, Tung

, Gilad

. RNA-seq: Impact of RNA degradation on transcript quantification. BMC Biol, 2014; 12:42.

33.

, Łabaj

, Li

, et al. A comprehensive assessment of RNA-seq accuracy, reproducibility and information content by the Sequencing Quality Control Consortium. Nat Biotechnol, 2014; 32:903–914.

34.

Cassavaugh

, Lounsbury

. Hypoxia-mediated biological control. J Cell Biochem, 2011; 112:735–744.

35.

Ratcliffe

PJ.

Oxygen sensing and hypoxia signalling pathways in animals: The implications of physiology for cancer. J Physiol, 2013; 591(Pt 8):2027–2042.

36.

Ziello

, Jovin

, Huang

. Hypoxia-Inducible Factor (HIF)-1 regulatory pathway and its potential for therapeutic intervention in malignancy and ischemia. Yale J Biol Med, 2007; 80:51–60.

37.

Ordys

, Launay

, Deighton

, McCulloch

, Whittle

. The role of mitochondria in glioma pathophysiology. Mol Neurobiol, 2010; 42:64–75.

38.

Oudard

, Boitier

, Miccoli

, Rousset

, Dutrillaux

, Poupon

. Gliomas are driven by glycolysis: Putative roles of hexokinase, oxidative phosphorylation and mitochondrial ultrastructure. Anticancer Res, 17:1903–1911.

39.

Heddleston

, Li

, Lathia

, Bao

, Hjelmeland

, Rich

. Hypoxia inducible factors in cancer stem cells. Br J Cancer, 2010; 102:789–795.

40.

Pestov

, Shcherbik

. Rapid cytoplasmic turnover of yeast ribosomes in response to rapamycin inhibition of TOR. Mol Cell Biol, 2012; 32:2135–2144.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.03 MB

0.00 MB

0.02 MB